RADIOGRAPHIC ASSESSMENT OF THE ASSOCIATION
OF UPPER FEMUR GEOMETRY AND TEXTURE
FEATURES TO HIP OSTEOARTHRITIS AND FRACTURE
Yaw Adjei
Master’s Thesis
Master’s Program in Biomedical Engineering: Biomechanics and Imaging
Faculty of Medicine
University of Oulu
2015
Adjei, Yaw (2015). Radiographic assessment of the association of upper femur geometry
and texture features to hip osteoarthritis and fracture. Research Unit of Medical Imaging,
Physics and Technology, University of Oulu, Master’s Thesis, 74 pages
Abstract
Osteoarthritis (OA) is a common joint disease found mostly in the elderly that
progressively leads to the loss of articular cartilage, along with subchondral bone changes
(sclerosis) and joint space narrowing, eventually resulting in joint failure. Osteoporosis
(OP), another bone disorder mostly of the elderly, is the gradual loss of bone tissue mass
(bone mineral density (BMD)) resulting in reduction in bone strength and an increased
risk of fracture. Bone geometry also plays a role in developing a fracture or not. OA
causes increase in bone volume, but fracture is still prevalent in OA patients. The aim of
this study was to investigate association of upper femur geometry and texture features to
hip OA and hip fracture using radiography.
Supine anteroposterior radiographs of hip and BMD from 125 postmenopausal women
were used for this research. Hip geometry parameters and texture related parameters were
obtained from the radiographs. Participants’ weight, height, and body mass index (BMI)
were also used in this research.
OA was found in women with higher weight, BMI, femur neck BMD, neck diameter
(ND), and neck cortex thickness, and shorter joint space width (JSW). OA women also
had higher homogeneity index of Laplacian images of femur neck and lower entropy of
Laplacian images of femur neck. Fractures were more common among women with lower
neck BMD, and higher femoral neck axis length, JSW, hip axis length (HAL) and
acetabular width (w). For cervical fractures, JSW and w were higher. Women with
trochanteric fractures had lower neck BMD and head diameter, and higher ND and HAL.
Upper femur geometry may play a role in the initiation and progression of OA and OP,
and trabecular microarchitectural changes in OA are relatively distinct. Higher weight,
BMI and neck BMD are predisposing factors for OA. Lower neck BMD is a predisposing
factor for trochanteric fracture.
Keywords: Osteoarthritis, osteoporosis, radiography, fracture, upper femur geometry,
image texture analysis.
List of Abbreviations and symbols
2D Two Dimensional
3D Three Dimensional
ADR Acetabulum Depth Ratio
BMD Bone Mineral Density
BMI Body Mass Index
CCD Charge-Coupled Device
CT Computed Tomography
d Acetabulum depth
ELV Extra Low Voltage
Ent_Lap_head Entropy, obtained from Laplacian of femur head
Ent_Lap_neck Entropy, obtained from Laplacian of femur neck
Ent_LBP_head Entropy, obtained from Local Binary Pattern of femur
head
Ent_LBP_neck Entropy, obtained from Local Binary Pattern of femur
neck
FACIT Fibril Associated Collagens with Interrupted Triple
helices
FNAL Femoral Neck Axis Length
FNC Femoral Neck Cortex Thickness
GLCM Gray Level Co-occurrence Matrix
HAL Hip Axis Length
HD Femoral Head Diameter
HI_Lap_head Homogeneity Index, obtained from Laplacian of femur
head
HI_Lap_neck Homogeneity Index, obtained from Laplacian of femur
neck
JSW Hip Joint Space Width
LBP Local Binary Pattern
MRI Magnetic Resonance Imaging
ND Femoral Neck Diameter
NICE National Institute for Health and Care Exchange
NSA Neck Shaft Angle
OF Obturator Foramen
SD Standard Deviation
SDM Signature Dissimilarity Measure
w Acetabulum width
µ Attenuation Coefficient
Contents
Abstract ............................................................................................................................. 1
List of Abbreviations and symbols ................................................................................... 2
Contents ............................................................................................................................ 4
1. Introduction ................................................................................................................ 6
2. Literature Review ...................................................................................................... 9
2.1 Articular cartilage, bone, and subchondral bone ................................................ 9
2.1.1 Composition and Structure of Articular Cartilage ...................................... 9
2.1.2 Composition and Structure of Bone .......................................................... 11
2.1.3 Composition and Structure of the Subchondral Bone ............................... 15
2.2 Osteoarthritis .................................................................................................... 16
2.2.1 Diagnosis of OA ........................................................................................ 17
2.2.2 Treatment of OA ....................................................................................... 19
2.3 Osteoporosis ..................................................................................................... 19
2.3.1 Diagnosis of Osteoporosis ........................................................................ 20
2.3.2 Treatment of OP ........................................................................................ 21
2.4 Fractures ........................................................................................................... 22
2.4.1 Causes of Fractures ................................................................................... 22
2.4.2 Diagnosis of Fractures............................................................................... 23
2.4.3 Hip Fractures ............................................................................................. 24
2.4.4 Treatment of Hip Fractures ....................................................................... 24
2.5 X-rays ............................................................................................................... 25
2.5.1 Basic Physics of X-rays ............................................................................ 25
2.5.2 X-ray tube.................................................................................................. 27
2.5.3 X-ray Detectors ......................................................................................... 28
2.5.4 X-ray Absorptiometry ............................................................................... 28
2.6 Hip Geometry ................................................................................................... 30
2.7 Trabecular Texture Analysis ............................................................................ 31
3. Aims of the Study .................................................................................................... 33
4. Materials and Methods............................................................................................. 34
4.1 Study Subjects .................................................................................................. 34
4.2 Imaging, OA grading, and measurement of geometrical parameters ............... 34
4.3 Evaluation of the Textural Parameters ............................................................. 37
4.4 Statistical Analysis ........................................................................................... 39
5. Results ...................................................................................................................... 40
6. Discussion ................................................................................................................ 47
7. Summary .................................................................................................................. 53
References ....................................................................................................................... 54
1. Introduction
Osteoarthritis (OA) and osteoporosis (OP) are two common medical conditions of the
elderly that are often associated with disability (or reduced mobility) (Lane 2007) and
fracture (Cosman et al. 2014, Hernlund et al. 2013, Kanis et al. 2013) respectively,
coupled with their huge economic burden they place on patients and the national health
care of many countries. At the advanced stages of the diseases most patients of OP or OA
may need a form of (replacement) surgery, especially if it involves loading bearing joints
like hips and knees.
OA probably results from the progressive degeneration of articular cartilage, along with
changes (sclerosis) of the subchondral bone. Other usual macroscopic changes of the
subchondral bone associated with OA are the formation of subchondral bone cysts and
osteophytes. (Buckwalter et al. 2005, Lane 2007, Li et al. 2013). OA generally affects all
tissues of the (synovial) joint; articular cartilage, subchondral and metaphyseal bone,
synovium, ligaments, joint capsules, and the muscles that acts across the joint, leading to
joint failure (Buckwalter et al. 2005). OA pathogenesis is not understood (Lane 2007,
Buckwalter et al. 2005). OA has no cure, but when detected early, the progression of the
disease can be impeded (Buckwalter et al. 2005, Egloff et al. 2012, Lane 2007).
Medical diagnosis of OA is usually confirmed with plain radiograph of the affected joint
based on the Kellgren-Lawrence (KL) grading with the presence of other complementary
symptoms of the disease. Radiographic diagnosis and grading of OA is based on joint
space narrowing, subchondral bone density increase, deformity of the bone ends, and
presence of osteophytes (Buckwalter and Martin 2006, Kellgren and Lawrence 1957). KL
grading is subjective, semi quantitative, with reported moderate to substantial intra- and
interrater reliability variation among assessors. (Damen 2014, Günther and Sun 1999, and
Kellgren and Lawrence 1957). Thus an improved diagnostic criteria of OA where there
will be higher specificity and reduced (or no) intra- and interrater differences, and user
independent will be a highly appreciated model.
Prior studies on radiographic categorization of OA mainly focuses on joint space width
measurement (Buckland-Wright 1999) and the presence of osteophytes. (Altman et al.
1991, Kellgren and Lawrence 1957) Kinds et al. (2011) demonstrated the use of bone
density estimation in discriminating OA from control patients from plain radiographs,
7
which is usually affected by image acquisition parameters and post-processing algorithms
(Kinds et al. 2011). Another potential method of extracting information related to bone
structure from plain radiograph is texture analysis. Texture analysis is independent of
imaging conditions. Hirvasniemi et al. (2014) showed the potential of Laplacian and
Local Binary Patterns (LBP) in bone OA analysis of the knee, whose effectiveness has
not been studied in hip OA as of now.
OP is a skeletal disease characterized by low bone mass and microarchitectural
deterioration of bone tissue, thereby predisposing a bone to fracture. (Cosman et al. 2014,
Hernlund et al. 2013, Kanis et al 2013, WHO 843 1994). OP is an asymptomatic disease,
and is usually detected in the advanced stage when the disease is complicated by
fracture(s)—spontaneous or by minimal trauma (Cosman et al. 2014).
Areal (or sometimes volumetric) bone mineral density (BMD) measurement, assessed
with Dual energy X-ray Absorptiometry (DXA), is the usual diagnostic procedure used
to confirm the existence of OP (Cummings et al. 1993, Johnell et al. 2005, Marshall et al.
1996, Schott et al. 1998). Notwithstanding the high specificity of BMD in predicting the
existence of OP, studies have shown that individuals with BMD outside the osteoporotic
range are also liable to fracture (Marshall et al. 1996, Schuit et al. 2004, Stone et al. 2003).
Biomechanical analysis also show that the mechanical strength of bone is not determined
by only BMD (bone mass), but also by other factors which are not captured in BMD
analysis. Pulkkinen et al. (2010) showed that the commonly used T-score criterion
classification of OP (T-score ≤ -2.5) better discriminates trochanteric fractures from
controls, and that about half of cervical hip fractures occur in individuals with BMD
outside the osteoporotic range. It has also been shown that individuals with trochanteric
fractures usually have general and higher bone loss (Duboeuf e al. 1997, Mautalen et al.
1996, Pulkkinen et al. 2004, Schott et al. 2005), whereas cervical hip fractures are dictated
by femur structural parameters (Duboeuf e al. 1997, Gnudi et al. 2002, Partanen et al.
2001, Pulkkinen et al. 2004, Pulkkinen et al. 2010, Pulkkinen et al. 2011). In order to
capture all the risk factors associated with developing fracture, other imaging modalities
like Computed Tomography (CT) and Magnetic Resonance Imaging (MRI) (Bauer et al.
2006, Baum et al. 2010, Herlidou et al. 2004, Huber et al. 2008, Hudelmaier et al 2005,
Lammentausta et al. 2006, Müller et al. 2006, Showalter et al. 2006), which are three-
dimensional (3D), have been used to assess the risk of fracture. With plain radiography
8
being relatively cheap and widely available, it still suffice to improve the specificity of it.
Skeletal trabecular texture (Benhamou et al. 2001, Boehm et al. 2009, Chappard C et al.
2005, Gregory et al. 2004, Vokes et al. 2006) and geometrical (Gnudi et al. 2002, Gnudi
et al. 2012, Partanen et al. 2001, Pulkkinen et al. 2004, Pulkkinen et al. 2010, Pulkkinen
et al. 2011) analyses of OP bones continue to be on the rise. In a recent study by Thevenot
et al. (2014), it was demonstrated that trabecular bone texture analysis was a better
predictor of femoral neck fracture than BMD using homogeneity index calculated from
gray level co-occurrence matrix (GLCM) of Laplacian images from plain radiographs.
In this study, the use of femur upper geometry and bone texture features in the prediction
of hip OA and fracture was investigated. Supine anteroposterior hip radiographs from 125
postmenopausal women were used for this research. BMD of each participant was also
taken. Hip geometry parameters and trabecular texture features were extracted from the
radiographs to ascertain their correlation with hip OA and fractures. Participants’ weight,
height, and body mass index (BMI) were also collected to assess their relevance in the
development of OA and fracture.
9
2. Literature Review
2.1 Articular cartilage, bone, and subchondral bone
The hip joint, which is a synovial joint, is made up of the following tissues; hyaline
articular cartilage, joint capsule, synovium, ligament, meniscus and subchondral bone.
The bone in the human hip joint include the upper femur and the lateral inferior section
of the pelvic (around the acetabulum). The hip transmits load from the trunk to the lower
limbs.
2.1.1 Composition and Structure of Articular Cartilage
Articular cartilage is a hyaline cartilage, with a substantial volume which does not ossify.
It consists of cells, water and macromolecule matrices. The matrices define the
mechanical properties of the articular cartilage. Chondrocytes, the only cell in normal
articular cartilage, contribute to about 1% of the total volume of mature human articular
cartilage. The wet weight of the articular cartilage is about 80% water. Articular cartilage
fluid also contains gases, cations (e.g. Sodium) and the macromolecules. Twenty to 40
per cent of the wet weight of the articular cartilage is structural macromolecules, i.e.
collagens, proteoglycans, glycoproteins, and non-collagenous proteins. (Buckwalter et
al. 2005).
Each chondrocyte is surrounded by an extracellular matrix, but remains metabolically
active (anaerobic) (Buckwalter and Mankin 1997), and have the essential organelles
required for matrix synthesis, including endoplasmic reticulum and Golgi membranes.
Dry cartilage contains about 60% collagen, 25 to 35% proteoglycans, and 15 to 20%
proteins and glycoproteins. Collagen type II is the most abundant collagen in the cartilage,
making up 90 to 95% of the collagens. The structural arrangement of the various collagen
fibers into a tight network throughout the articular cartilage defines the tensile stiffness
and strength of the tissue, and also helps in holding together the large proteoglycans and
the non-collagenous proteins. The principal proteoglycans found in articular cartilage are
large aggregating proteoglycan monomers or aggrecans and small proteoglycans
including decorin, biglycan, and fibromodulin (Buckwalter and Mankin 1997). The
10
proteoglycans are usually negatively charged, and are bonded to the cations in the
cartilage fluid. Most of the interfibrillar space of the cartilage matrix are filled by
aggrecans, making up about 90% of the total cartilage proteoglycan mass. There are a
variety of non-collagenous proteins and glycoproteins found in normal articular cartilage,
but their functions and behaviour are not well understood. (Buckwalter et al. 2005).
The depth from the surface of the articular cartilage determines its cellular morphology
and probably function. The depth also determines the composition, organization,
mechanical properties of the matrix. The articular cartilage can be divided into four zones;
calcified, deep, transitional and superficial zones. Figure 1 shows the structure of articular
cartilage and subchondral bone, with the appropriate zones. The boundaries of the various
zones cannot be distinctively defined, as there is a smooth transition of cellular and matrix
composition from one zone to the next zone. (Buckwalter et al. 2005).
The superficial zone is the thinnest zone of the articular cartilage, and it is made up of
two layers. An acellular sheet of fine fibrils with little polysaccharide is the topmost layer
of the superficial zone. Lying inferiorly to the acellular sheet is a flat ellipsoidal-shaped
chondrocyte whose major axis are parallel to the articular surface. The matrix synthesized
by these chondrocytes has a high concentration of collagen, but low concentration of
proteoglycan relative to the other cartilage zones. The high concentration of collagen in
this layer contribute to the relatively high tensile stiffness and strength, but low
permeability of the superficial zone. This zone also contains the highest concentration of
water and fibronectin. (Buckwalter et al. 2005).
The transitional zone is usually several times bigger, and its cell have higher
concentration of synthetic organelles, endoplasmic reticulum and Golgi membranes
compared to the superficial zone. Cells of the transitional zone appear spheroidal in shape
and synthesize a matrix that has larger-diameter collagen fibrils, a higher concentration
of proteoglycan, but lower water concentrations and collagen than superficial zone
matrix. (Buckwalter et al. 2005). The transitional zone is very resistant to compression
(kneejointsurgery.com 2015).
The deep zone has spheroidal shaped chondrocytes, which lie perpendicular to the joint
surface. The largest-diameter collagen fibrils, and the highest concentration of
proteoglycans are found in this zone, but has the least water concentration. (Buckwalter
et al. 2005). The deep zone is also very resistant to compression. (kneejointsurgery.com
11
2015). The calcified zone, which is relatively thin, separates the uncalcified deep zone
from the subchondral bone. The calcified zone cells have a smaller volume relative to the
deep zone, with only small amounts of endoplasmic reticulum and Golgi membranes.
(Buckwalter et al. 2005). A thin basophilic line, known as tidemark, sectionalizes the
calcified zone from the uncalcified zones (Buckwalter et al. 2005), and a sharp borderline
called cement line also separates the calcified cartilage from the subchondral bone (Li et
al. 2013).
Figure 1. The Structure of Articular Cartilage and Subchondral Bone.
2.1.2 Composition and Structure of Bone
The human bone is a metabolically active live tissue, undergoing a lifetime modelling
and remodelling. Bone modelling is the process whereby bones change their overall shape
and size in response to physiological influences and mechanical forces to help the
skeleton to adapt to changing biomechanical forces so as to remove damage and maintain
Subchondral Bone
Collagen Fibrils
Zone 1: Superficial Zone
Trabecular Bone
Zone 4:
Calcified Zone
Zone 2:
Transitional Zone
Zone 3:
Deep(Radial) Zone
Articular Surface
Chondrocyte
Ground Substance
Trabecular Bone
Subchondral Bone
Tide Mark
Cement Line
12
bone strength (Clarke 2008, Kobayashi et al. 2003). Bone remodelling is the process in
which old and micro-damaged bones are replaced with new mechanically stronger bones
to help preserve bone strength (Burr 2002, Clarke 2008, Parfitt 2002, Raisz 1999). Bone
modelling rarely occurs in adults than remodelling (Kobayashi et al. 2003), but may be
altered by diseases or pharmacological agents (Lindsay 2006, Ubara et al. 2003, Ubara et
al. 2005). A variety of human bones make-up the skeletal system, the classification of
which may be based on macrostructure, or microstructure and content. At the
macrostructure level, a bone may be classified as, long, short, flat or irregular. By content,
a bone may be trabecular or cortical.
Long bones (Figure 2) are made-up of a hollow shaft or diaphysis; a flared, cone shaped
metaphysis beneath the growth plates; and above the growth plate is rounded epiphysis.
A dense and solid cortical bone make-up the diaphysis, while the metaphysis and
epiphysis are composed of honeycomb-like network trabeculae bone surrounded by a
relatively thin shell of a dense cortical bone. At joints, subchondral bone surrounds the
trabecular bone, which are finally lined by articular cartilage. (Clarke 2008).
In human adults, 80% of the skeleton is cortical bone, with the remaining 20% being
trabecular bone, and the ratio of cortical to trabecular is different for different bones and
skeletal sites. The radial diaphysis has the highest amount of cortical bone, 95% to 5%
trabecular. (Clarke 2008).
Osteons are the basic units of both cortical and trabecular bones. Cortical osteons are
known as the Haversian system, which are in the shape of a cylinder, with approximate
length of 400mm and 200mm base width, and forming a network of branches within the
cortical bone. Haversian system walls are formed from concentric lamellae.
Metabolically, cortical bones are less active relative to trabecular bone in humans. The
higher the cortex undergo remodelling, the less dense it becomes, thus increased porosity.
Normally, healthy aging adults have a thinning cortex and increased cortical porosity.
(Clarke 2008).
Lining the inner and outer surfaces of cortical bones are the endosteum and periosteum
respectively. The periosteal surface normally experience higher bone formation than
resorption, which contributes to increase in bone diameter with age. However, endosteal
bone formation is typically less than resorption, contributing to the expansion of marrow
13
with age. Osteons of the trabecular are called packets, which are made up of plates and
rods averaging 50 to 400mm in thickness. (Clarke 2008).
Cortical and trabecular bones are normally formed in lamellar pattern, which involves the
laying of collagen fibrils in alternating orientations (Eriksen et al. 1994, Seeley et al.
1998), which contribute to the strength of normal lamellae bones. Woven bones, which
are made of collagen fibril laid in a disorganised manner, are weaker than lamellae.
Woven or primary bone is the first type of developmental bone formed, and may also be
seen in high bone turnover abnormal conditions such as osteitis fibrosa systica and Paget’s
disease. (Clarke 2008).
Surrounding the outer cortical bone surface is the periosteum, a fibrous connective tissue
containing blood vessels, nerve fibres, and osteoblasts and osteoclasts. At the outer
cortical surfaces of the bone, the periosteum perforate into the underlying bone tissues,
hence attaching tightly to the bone by thick collagenous fibers known as the Sharpey
fibres (Clarke 2008, Seeley et al. 1998).
During childhood and adolescence, longitudinal and radial growth of the bones occurs.
During bone remodelling, discrete packets of old bones are removed and replaced by
newly synthesized proteinaceous matrix, which subsequently mineralized to form new
bone, thus preventing bone microdamage. Bone remodelling starts during foetal
development till death. (Clarke 2008).
Four types of cells are found in human bone; osteoprogenitor cells, osteoblasts,
osteocytes, and osteoclasts. Osteoblasts form bone tissues, osteocytes maintain them, and
are broken down by osteoclasts. Osteoprogenitor cells are precursors to osteoblasts,
which become osteocytes after matrix mineralization. Osteoprogenitor cells, derived from
mesenchyme, are unspecialised and divide through mitosis, and are found on all bone
surfaces. (University of Glasgow 2015).
Terminally differentiated osteoblasts are known as osteocytes, and osteocytes function
within syncytial networks to support bone structure through intercellular adhesion, and
bone metabolism by regulating the exchange of mineral in the bone fluid within lacunae
and canalicular network. During osteolysis (bone resorption), osteocytes may function as
phagocytes due to their lysosome content. Osteocyte-osteoblast cell syncytium are
primarily mechanosensors (Bonewald 2006) and that stress signals from bending and
stretching of bone are transduced into biological activity by them. The average half-life
14
of osteocytes is 25 years (Knothe Tate et al. 2004). Oestrogen deficiency (Plotkin et al.
2005) and glucocorticoid treatment may trigger osteocyte apoptosis. Bone remodelling is
Figure 2. Structure of Normal Bone.
carried out by bone remodelling units, composed of tightly coupled group of osteoclasts
and osteoblasts that sequentially resorb old bones and form new ones. Osteoclasts secret
acidifying H+ ions to dissolve bone matrix. Osteoblasts that are matured and active secrete
collagen type I and other matrix proteins in a vectorial fashion to form bone. These
matured active osteoblasts have large nuclei, enlarged Golgi structures, and extensive
endoplasmic reticulum. Microarchitectural deterioration is caused by high bone turnover
with greater bone resorption than formation. (Clarke 2008).
At the molecular level, composition of bone by percentage is; 50 to 70% mineral, 20 to
40% organic matrix (protein), 5 to 10% water, and less than 3% lipids. Hydroxyapatite
(Ca10 (PO4)6 (OH) 2) is the dominant mineral in bones, with small amounts of Carbonate,
Magnesium and acid Phosphate. Collagenous proteins make up 85 to 95% of bone
Articular Cartilage
Subchondral Bone
Trabecular
Bone
Epiphyseal Plate
Epiphysis
(Proxiaml)
Metaphysis
Cortical
Bone
Diaiphysis Endosteum
Periosteum
15
proteins, mostly collagen type I (Brodsky and Persikov 2005), with very small amounts
of collagen types III and V, and Fibril Associated Collagens with Interrupted Triple
helices (FACIT). Collagen types III, V, and FACIT are only found in bones at certain
stages of bone formation, and may help determine the diameter of collagen fibril. The
nonfibrillar collagens, FACIT, act as molecular bridges to facilitate the organisation and
stability of bone extracellular matrices. The remaining 10 to 15% bone protein, which are
non-collagenous, include proteoglycans, glycosylated proteins, and γ-Carbonated
Proteins. On molar basis, equal amounts of collagenous and non-collagenous proteins are
synthesized and secreted by osteoblasts. (Clarke 2008).
Bone’s mechanical rigidity and load-bearing strength is determined by its mineral
content, while its elasticity and flexibility is dependent on the amount of organic matrix
present (Seeley et al. 1998). Laying of bone mineral first begins at the hole zones between
the ends of collagen fibrils. Vitamin D indirectly excites the mineralization on
unmineralized bone matrix. Dentin matrix protein 1 and bone sialoprotein are the main
promoters of bone mineralization. (Clarke 2008).
About 50 to 70% of bone strength is determined by bone mass. The larger a bone is, the
stronger it is regardless its mineral density. Bone strength increases to the fourth power
for every radial increase. The ratio of cortical to trabecular bone at a given skeletal site
independently affect bone strength. (Clarke 2008).
The skeletal bones serve as a framework for the rest of the body, aid in movement and
locomotion through the provision of levers for the muscles, maintain mineral homeostasis
and acid-base balance, maintain a conducive environment for haematopoiesis within the
marrow spaces (Parfitt 2002), store growth factors and cytokines for future use, and
enclose and protect vital internal organs and structures. (Clarke 2008, Seeley et al. 1998).
2.1.3 Composition and Structure of the Subchondral Bone
The subchondral bone lies just underneath the deep calcified articular cartilage (Burr and
Gallant 2012, Madry et al. 2010). The subchondral bone consist of two distinct anatomic
entities: the subchondral bone plate and subchondral trabecular bone (Clark and Huber
1990), both of which are lamellae. (Goldring and Goldring 2010, Li et al. 2013). The
subchondral bone plate, a thin cortical lamella, lies just beneath the calcified cartilage
16
(Clark and Huber 1990, Milz and Putz 1994). The subchondral bone (end) plate is porous
structure with channels occupied by vessels and nerves. These vessels and nerves run
from the subchondral trabecular to the calcified cartilage (Holmdahl and Ingelmark 1950,
Madry et al. 2010). The channels are more at highly stressed areas of a joint. The channels
are narrower and form a tree-like mesh in regions with thicker subchondral bone plate,
while they are wider and ampulla like at thin regions (Milz and Putz 1994).
Supporting the subchondral bone plate is the subchondral trabecular bone (Clark and
Huber 1990, Madry et al. 2010). The subchondral bone trabecular functions as a shock-
absorber (Kawcak et al. 2001) and also may play important role in cartilage nutrient
supply and metabolism (Castañeda et al. 2012). The subchondral trabecular is more
porous and metabolically active, with more blood vessels, sensory nerves, and bone
marrow than the subchondral bone plate (Suri and Walsh 2012). Subchondral trabecular
bone structure exhibit preferential spatial orientation and parallelism, that is,
mechanically anisotropic (Holopainen et al. 2008). The subchondral bone is a dynamic
structure that adapt to mechanical stress imposed across the joint, thereby adjusting
trabecular orientation and scale parameters in according to the principal stress on the joint
through bone modelling and remodelling (Goldring 2012, Holopainen et al. 2008,
Kawcak et al. 2001, Milz and Putz 1994, Walker et al. 2013).
2.2 Osteoarthritis
OA is a joint disease characterize primarily by the progressive loss of normal articular
cartilage structure and function due to cartilage degeneration, along with subchondral
bone remodelling and sclerosis, and usually subchondral bone cysts and marginal
osteophytes formation. OA causes all the joint structures to undergo pathological change
simultaneously, but usually the first to be affected is the superficial zone of the articular
cartilage. (Li et al. 2013). Other names for OA are degenerative arthritis and hypertrophic
arthritis. OA is one of the leading causes of disability in the elderly (Lane 2007). OA is
more common in people over 60 years, and in elderly women more than men (Li et al.
2013), and it usually leads to inflammation of the affected joint, (Lane 2007) hence the
suffix –itis. OA usually affect the vertebrae, hip, knee, foot and hand joints. In OA, the
progressive loss of articular cartilage occur together with the attempted repair of articular
17
cartilage, remodelling and sclerosis of the subchondral bone, and usually the formation
of subchondral bone cysts and marginal osteophytes (Lane 2007, Li et al. 2013). Deeper
trabecular texture analysis has also revealed microscopic changes in the trabecular as well
(Papaloucas et al. 2005). In some other instances, especially following extreme
mechanical damage, chondrocytes of the articular cartilage may die, making tissue repair
almost impossible. OA may also cause pain, restriction of movement, (Lane 2007)
crepitus with movement, effusion and deformity at the affected joint. Changes that occur
in the subchondral bone include increased density or subchondral sclerosis, formation of
cysts and fibrous or cartilaginous tissue, and osteophytes. (Lane 2007, Li et al. 2013). At
the latter stages of the disease, when articular cartilage has been completely lost,
thickened, dense subchondral bone will be articulating with a similar opposing denuded
bony surface, which is usually the cause of crepitus. OA may lead to shortening of the
limb (s) involved, deformity and instability due to the combined effect of bone
remodelling and articular cartilage loss. Articular cartilage loss leads to secondary
changes in the synovium, ligaments, capsules and the muscles that move the affected
joint. The synovial membrane usually become inflamed in response to the change of its
normal environment. (Buckwalter et al. 2005).
OA may develop of an unknown cause, for which it is called primary osteoarthritis or
idiopathic osteoarthritis (Lane 2007), which happens to be the most common
osteoarthritis. However, osteoarthritis caused by joint injury, infection, hereditary factor
(s), or developmental, metabolic (Lane 2007) and neurologic disorders is known as
secondary osteoarthritis. (Buckwalter et al. 2005).
Postulated risk factors for primary OA risk are advanced age, genetic predisposition for
OA, (Lane 2007) hormonal and metabolic disorders, inflammation, and immunologic
disorders. (Buckwalter et al. 2005).
2.2.1 Diagnosis of OA
OA diagnosis primarily includes symptoms identification, review of medical history, and
physical examination. Plain radiography is usually used to ascertain the existence of OA.
Magnetic Resonance Imaging (MRI), arthroscopy, and laboratory examination of
synovial fluid are other complementary procedures in the diagnosis of OA.
18
Symptoms and signs of OA include, gradual onset of pain, morning stiffness and pain
(Lane 2007), restriction of movement, crepitus with joint movement, joint effusions, joint
deformities and subluxations (Buckwalter et al. 2005). OA diagnosis with plain
radiographs is usually based on the Kellgren-Lawrence grading (Table 1) (Kellgren and
Lawrence 1957).
Table 1. Classification criteria for different Kellgren-Lawrence (KL) grades in the hip.
MRI, a non-invasive imaging modality with no radiation exposure can be used to
visualize joint morphology; volume, thickness, articular cartilage, curvature, and changes
in subchondral bone (Burstein and Gray 2003, Lammentausta et al. 2006). A white blood
cell count of less than 1000 per cubic millimetre is an indication of OA. Alternatively,
erythrocyte sedimentation rate of less than 20mm/hr is also an indication of OA (Lane
2007).
According to the American College of Rheumatology hip OA is expected if hip pain exist
with at least any 2 of the following;
an erythrocyte sedimentation rate of less than 20mm/hr
radiographic evidence of femoral or acetabular osteophytes
radiographic evidence of joint space narrowing (superior, axial, or medial)
(Altman et al. 1991).
KL grade Description
0 – Normal No radiographic features of osteoarthritis (OA)
Doubtful narrowing of joint space and possible osteophytes 1 – Doubtful OA
2 – Minimal OA Definite osteophytes and narrowing of the joint space
3 – Moderate OA Moderate multiple osteophytes, definite narrowing of the joint
space, some sclerosis, and possible deformity of bone contour
4 – Severe OA Large osteophytes, marked narrowing of the joint space, severe
sclerosis, and definite deformity of bone contour
19
2.2.2 Treatment of OA
OA treatment therapy aim to relieve pain and preserve function (Buckwalter et al. 2005,
Egloff et al. 2012, Lane 2007). Both Pharmacological and non-Pharmacological regimens
are used in the treatment of OA (Lane 2007). Fransen et al. (2014) showed that exercise
interventions is an effective treatment for OA of the hip.
Randomised studies on the treatment of OA using aquatic therapy and acupuncture
showed functional improvement (Hinman et al. 2007, Witt et al. 2006). Weight loss may
also be a good step in managing OA (Lane 2007). Walking with cane to improve balance
and the use of insoles have also been suggested to limit the progression of OA.
Pharmacological interventions like analgesics, at least effective doses, may be prescribed
to relief pain. Analgesics like Nonsteroidal Anti-inflammatory Drugs (NSAIDS) can
cause gastrointestinal disturbance like peptic ulcer, hence it is usually taken with proton-
pump inhibitors. (Lane 2007).
For OA patients experiencing chronic discomfort and substantial functional impairment,
arthroplasty, osteotomy, or at worst total (hip) joint replacement surgery may be
considered to reduce pain and disability (Egloff et al. 2012, Lane 2007).
2.3 Osteoporosis
Osteoporosis (OP), literally meaning ‘porous bone’ is a disease of the skeletal system
characterised by low bone mass and microarchitectural deterioration of bone tissue, which
predisposes a bone to fracture (Cosman et al. 2014, Hernlund et al. 2013, WHO 843
1994). Sites usually affected by OP are the spine, hip, distal forearm and proximal
humerus. OP fractures may remain asymptomatic or cause pain, immobility (especially
hip fractures), loss of function (especially vertebra fractures) and mortality as well. The
human bone is in a state of continual remodelling, with older bone tissues being replaced
by new ones. A disruption in the balance between the rates of bone replacement and
degradation, specifically with higher degradation rate than replacement, leads to
osteoporosis (Cosman et al. 2014). OP is diagnosed by measuring the BMD (Blake and
Fogelman 2007, Davidsson 2010), which is the amount of bone (mineral) mass per unit
volume (volumetric density) or per unit area (areal density) (Davidsson 2010), both of
20
which can be measured in vivo using densitometric techniques (Blake and Fogelman
2007, Davidsson 2010). DXA remains the most widely used densitometric technique for
measuring BMD. Other common methods of measuring BMD are Quantitative
Ultrasound (QUS) and Quantitative Computed Tomography (QCT) (Davidsson 2010).
DXA scans can be used to assess the bone mineral content (BMC) and BMD of the whole
skeleton as well as of specific sites, which may be sites most vulnerable to fracture
(Davidsson 2010, Genant et al. 1996). With most conventional X-rays being two
dimensional, areal density rather than true volumetric density is used. In vitro studies of
isolated vertebra and proximal femur showed that areal BMD accounts for about two-
thirds of the variance of bone strength. The vertebral column and the femur remains the
most assessed areas for BMD measurements. (Kanis et al. 2013).
X-ray-based BMD measurements exposes the body to (small amounts of) radiation.
2.3.1 Diagnosis of Osteoporosis
The most often used parameter for the description of BMD is the T- or Z-score, both of
which are measured in units of standard deviation (SD). A person’s T-score is the
difference between his/her BMD and the mean BMD in a healthy young adult, of the
same gender and ethnic group, with the difference expressed as a ratio of the young adult
population standard deviation (Figure 3). In Z-scores, a person’s BMD is subtracted by
the mean BMD of persons of similar age, gender and ethnic group, and expressing the
difference relative to age-matched population standard deviation. (Blake and Fogelman
2007) Z-score is mostly used in the assessment of children and adolescents OP (Kanis et
al. 2013). Operationally, OP is defined based on the T-score for BMD assessed at the
femoral neck. A person is classified as being osteoporotic if his/her BMD is 2.5SD or
more below the mean BMD of a young adult of the population in question (WHO 2004)
(Figure 3). A low BMD with SD between -1 and -2.5 is referred to as osteopenia. This
diagnostic criteria is used because for any age and BMD at the femoral neck and hip,
major osteoporotic fracture risks are the same for men and women (Looker et al. 1998),
thus the femoral neck has been adopted as the reference site for the diagnosis of
osteoporosis (Kanis and Glüer 2000). OP fracture risk factor of at least 1.5-fold exist for
each SD decrease in BMD from the mean (Carmona 2004, Marshall et al. 1996). Some
21
risk factors for developing OP are age, glucocorticoid exposure, rheumatoid arthritis and
low calcium intake (Cosman et al. 2014). Factors that predisposes a person to OP include:
lifestyle (e.g. smoking, alcohol abuse, frequent falling), genetically predispose disease
(e.g. cystic fibrosis), endocrine disorders (e.g. diabetes), rheumatologic and autoimmune
diseases (e.g. lupus and rheumatoid arthritis), and so on (Carmona 2004).
Figure 3. BMD Classification using T-scores.
2.3.2 Treatment of OP
A variety of lifestyle treatment options exist for persons suffering from OP. These include
adequate intake of Calcium and Vitamin D, lifelong participation in regular weight-
bearing and muscle strengthening exercise, cessation of tobacco use for smokers,
identification and treatment of alcoholism for alcoholics, and treatment of risk factors for
falling. (Cosman et al. 2014).
For persons at higher risk of fracture (very low BMD), pharmacological interventions,
and usually in combination with lifestyle treatment options, may be considered.
Treatment with pharmacological interventions should not be a lifelong option, as there
are usually unpleasant side effects, of which some are life threatening. Some of the effects
of pharmacological drugs are, oesophagus and stomach inflammation, renal function
50
15
Osteoporosis Osteopenia Normal
4 3 2 1 0 -1 -2 -3 -4
Standard Deviation (SD) units (or T-score)
Percentage of Population (%)
22
disturbances, rhinitis, epistaxis, myocardial infarction, stroke, invasive breast cancer,
pulmonary emboli, deep vein emboli and osteosarcoma. (Cosman et al. 2014).
2.4 Fractures
Fracture is a break in the continuity of a bone, with or without displacement of fragment.
Fracture is always accompanied by soft tissue damage—torn vessels, bruised muscles,
lacerated periosteum, and contused nerves (Duckworth and Blundell 2010). Complete
fractures have complete break through the bone radial cross-section at the point of fracture
and complete separation of bone is involved. In incomplete fractures, break in bone
continuity is not complete radially at the site of fracture, but bone is continuous at parts
of the bone at the point of fracture, e.g. greenstick fracture. (Newton 2015).
Non-comminuted fractures results in only two segments of bone, whereas comminuted
fractures results in more than two bone segments. Comminuted fractures are usually
difficult to fix compared to non-comminuted fractures, especially gross comminuted
fractures. (Duckworth and Blundell 2010, Newton 2015).
Fracture that remain within the skin and the musculature surroundings is known as closed
fracture. On the other hand, a fracture that is exposed to outside environment is known as
open fracture. (Newton 2015).
Fractures may also be classified according to the anatomical location of the fracture.
Examples of such classification are, diaphyseal, metaphyseal, epiphyseal plate,
epiphyseal, condular, and articular fractures. In addition, the exact bone involved helps
in ease identification of the location of the fracture (Newton 2015).
Alternatively, fractures may be named according the magnitude and direction of the force
causing it. The common classifications are; transverse, spiral or oblique, greenstick,
crush, burst, avulsion, and subluxation. (Duckworth and Blundell 2010, Newton 2015).
2.4.1 Causes of Fractures
The underlying cause of a fracture may be pathological or non-pathological (trauma).
Trauma remains the major cause of fractures, which may be e.g. as a result of automobile
or sports injury. Traumatic fractures are rarely predictable, since the amount of force
23
causing it is rarely calibrated. Most traumatic fractures are comminuted (or multiple).
Pathological fractures are caused by an underlying bony or systemic disease (s), that
makes one, many, or all bones of the skeletal system abnormal, thus becoming more
susceptible to fracture. When bone becomes susceptible to fracture, a low-energy trauma
usually fractures such bones. Even the person’s own weight may necessitate a pathologic
fracture, thus fracture occurs spontaneously without any major trauma. Common diseases
causing pathological fracture are type 2 diabetes (Giangregorio et al. 2012, Schwartz et
al. 2011), neoplasia, bone cysts, osteoporosis, nutritional hyperparathyroidism, and
osteomyelitis. (Newton 2015).
FRAX©, a computer based tool developed by a WHO Scientific group is also used in
many health centres to predict probability of a future fracture. According to FRAX©,
factors that predispose a person to fracture are; age, gender, prior osteoporotic fracture,
low femoral neck BMD, low body mass index (BMI), oral glucocorticoids intake,
rheumatoid arthritis, parental history of hip fracture, smoking, excessive alcohol intake,
and other secondary causes of osteoporosis. (Cosman et al. 2014, University of Sheffield
2015).
2.4.2 Diagnosis of Fractures
Fractures diagnosis is done through review of fracture history, physical signs and
radiological examination. Knowledge about the trauma type and severity may reveal an
underlying pathological condition or not. Common symptoms like pain (and the severity
of it), and loss of function are important indications of the type, location, and cause of
fracture. Loss of sensation or motor power is an indicative of nerve or vascular damage.
Physical indicators like tenderness, deformity, swelling, local temperature increase,
crepitus, and loss of function, which may be present or not, are also associated with
fractures. (Duckworth and Blundell 2010, Newton 2015).
X-ray examination gives additional information to that obtained by clinical judgment
described above. Plain radiographs, which is two-dimensional (2D) is usually used.
(Newton 2015) CT may be used to obtain 3D image of the fractured bone. Radiography
may complement physical monitoring and evaluation of fracture healing process.
24
Radiography also helps reduce incision/open area during surgery. (Duckworth and
Blundell 2010).
2.4.3 Hip Fractures
Fractures occurring in the area between the edge of the femoral head and 5cm below the
lesser trochanter are termed as hip fractures (NICE 2014). Hip fractures are generally
classified into two main groups, based on their location with regard to the capsule—
intracapsular, and extracapsular fracture (Frost et al. 2010, Jones et al. 2015). Subgroups
also exist within these two main classifications, depending on fracture location, fracture
level, displacement or not, and comminution (Jones et al. 2015).
Fracture of bones within the hip capsule are termed intracapsular, and the most general
types of them are femoral head fracture and femoral neck (cervical) fracture. Cervical
fractures are of great medical concern because of their tendency to damage the small
intracapsular vessels that provide the majority of the blood supply to the femoral head,
thereby leading to higher morbidity and mortality from cervical fractures (Jones et al.
2015).
Extracapsular fractures are fractures occurring outside the hip capsule, and are commonly
referred to as trochanteric fractures (Jones et al. 2015, NICE 2014).
2.4.4 Treatment of Hip Fractures
Fracture treatment generally involve reduction, maintenance of reduction, and
rehabilitation. Open fractures require emergent medical attention, and often debridement
may be needed. (Duckworth and Blundell 2010).
Fracture reduction is the surgical restoration of a fractured dislocated bone to the normal
bone alignment. Some (closed) fractures may be reduced by manipulation under
anaesthesia. Where anaesthesia is inappropriate, traction (usually for trochanteric
fractures), a weight attached to gently pull bones together, may be used (Hip Fracture
2015). Misalignment of bone (fragment) may lead to functional impairment, difficulty in
mobility or immobility, difficulty in uniting bones, soft tissue (nerves and vessels)
damage, and low aesthetic appearance. Sometimes open reduction is employed for
25
accurate bone fixation. Fractured bone may also be stabilized with a nail or fixator after
traction. (Duckworth and Blundell 2010).
After reduction, stability is maintained intrinsically for fractures which do not require
additional stabilization, otherwise external splintage or internal fixation is applied
(Duckworth and Blundell 2010). Most hip fractures are kept stable using internal fixation
devices like screws, pins, plates (fastened with screws) and nails. Some displaced hip
fractures may require hip arthroplasty (hemiarthroplasty or total hip replacement) (Hip
Fracture 2015, NICE 2014).
In rehabilitation, the affected limb is moved and used as much as possible, and it is done
when there is cortical-to-cortical union. Rehabilitation usually helps to stimulate
complete union and prevent joint stiffness. (Duckworth and Blundell 2010).
2.5 X-rays
2.5.1 Basic Physics of X-rays
X-rays are emitted when high speed electrons are stopped by a target metal. The most
common means of producing electrons is by thermionic emission. In thermionic emission,
a current is passed through a metal of high melting point, where the current (energy)
gained by the metal (free) electrons is converted to electron vibrational energy, causing
the metal to heat. The vibrational energy of the electrons causes some of the electrons to
move to higher energy state, thus becoming loosely bound to the metal atoms (Tutorvista
2015). With a higher positive voltage (metal) nearby the heated metal, the loosely bound
electrons escape their atoms toward the higher positive voltage metal. The higher voltage
metal is referred to as anode (target), and the heated metal is the cathode. (A level Medical
Physics 2013, Beiser 2003a).
The accelerated electrons, upon reaching the anode with sufficient energy is usually able
to knock-out electrons from the anode metal atoms, and give it sufficient kinetic energy
to escape the metal atom (A level Medical Physics 2013). The minimum amount of energy
required to liberate an electron from the surface of a metal is known as the work function
of the metal (Beiser 2003a). The knocking out of electrons from the anode leads to the
creation of holes, thereby ionizing the metal anode. If the accelerated electron from the
26
cathode has much energy, inner electron is liberated from the target anode atom. With
inner electron liberation, the next higher orbital electron drops to fill the hole created in
the lower orbital. The de-excitation of the higher orbital electron to lower orbital leads to
release of photon whose energy is equal to the difference in energy level between the
higher and lower orbital.
The photon released by the de-excitation of atom electrons is known as X-ray, for its
frequency is in the x-ray range of the electromagnetic spectrum, with frequencies just
above the Ultraviolet and below Gamma rays. Estimated frequencies of x-rays are 1017 -
1020 Hz.
The faster the accelerated electron (high kinetic energy), the more penetrating the
resulting X-rays, and the greater the number of accelerated electrons, the greater the
intensity of the resulting X-ray photon. For heavier atoms like tungsten, upon knocking
out of an inner electron, e.g. a K-shell electron, a series of electron de-excitation happen,
due to the subsequent filling of the lower orbital by higher orbital (s) till the hole moves
to the last (outermost) orbital, all of which produces radiation of different wavelengths
due to the difference in energy between the higher and the lower orbitals involved. This
explains why X-ray spectra of metals are continuous in nature, with the highest intensity
corresponding to the energy difference between L- to K-shell orbitals electron transition.
(Beiser 2003a).
Two main types of X-ray radiation are produced using thermionic emission;
bremsstrahlung (braking radiation) and characteristic radiation. Bremsstrahlung is
caused by the decrease in kinetic energy of the accelerated cathode electrons by the target
anode atoms’ electric field as the electrons get closer to the anode, and also causing a
change in trajectory of the accelerated electrons. This change in kinetic energy is
converted to X-ray photon radiation. The greater the kinetic energy of the accelerated
electron and the heavier the anode metal, the more energetic the bremsstrahlung.
Characteristic radiation is emitted when K-shell electron is knock-out by the accelerated
electron and subsequent filling by higher orbital electron (s). For multi electron (heavier)
metals like Molybdenum, there could be more than one characteristic radiation, since not
only L-K electron transitions occur after knocking out K-shell electrons. (Beiser 2003a,
Nave 2015).
27
Another phenomenon that contribute to the continuum of x-ray spectra in the Auger
phenomenon. In the Auger phenomenon, higher energy x-ray photons emitted due to
electron transition eject other higher orbital electrons of the same atom of the anode metal,
leading to a decrease in the original x-ray photon energy that ejected the electron. (Beiser
2003b).
2.5.2 X-ray tube
The X-ray tube (figure 4) is an evacuated (vacuum) glass tube with a heating cathode and
target anode some distance apart, and maintained at several kilovolts potential difference.
Electrons are generated from the cathode through thermionic emission. The tube is
evacuated to allow the flow of electrons from the cathode to the anode unimpeded. The
anode is tilted to direct the generated x-ray through a relatively thin section of the glass
envelope. The tilt of the target results in glancing collision by the electrons, with most of
their energy (about 99%) going into heat. Thus effective cooling mechanism of the anode
of an x-ray tube is required. This is achieved by rotating the anode, with the end immersed
in a flowing water or oil. (Beiser 2003a).
Figure 4. X-ray Tube.
0V
x-ray beam
window
vacuum tube
cathode
filamen
t
anode
ELV x-ray beam
kV
accelerated
electrons
ELV = Extra Low Voltage
28
2.5.3 X-ray Detectors
A variety of x-ray detectors are available, they can be broadly be classified, according to
the use, as for; imaging and dose measurement (Hendee and Ritenour 2002b and 2002c).
For imaging, image contrast results from the difference in attenuation coefficients of the
various body tissues (Hendee and Ritenour 2002a). Dense tissues like bones have high
attenuation coefficients and as such appear brighter due to low energy X-ray hitting the
detector, whereas loose or low dense tissues like fat and lipids have low attenuation
coefficients and as such appear darker due to high energy x-rays hitting the detector
(Hendee and Ritenour 2002c). Most imaging detectors use scintillator, e.g. Thallium-
doped Caesium Iodide ((CsI) Tl) and Sodium Iodide (NaI (Tl)), which convert X-ray
photons to light photons at an amplification, thus reducing the radiation exposure to the
patient (Hendee and Ritenour 2002b). The needed image is then produced, on a film (e.g.
silver bromide and silver iodide crystals in gelation matrix), using computed radiography
technology (image first captured on storage phosphor, and later released upon exposure
to light through a photomultiplier, and a computer is used to combine the received signals
into image), or using semiconductor technology to convert the light photons to electric
signals (photodiode/charge-coupled device (CCD)) which may then be processed with a
computer to produce the needed image in digital format (Hendee and Ritenour 2002c).
Most DXA systems use digital imaging technology because of their high spatial
resolution (Davidsson 2010).
2.5.4 X-ray Absorptiometry
X-ray used in diagnostic radiology interaction with matter (body tissue) results in one of
these three processes: coherent (elastic or Rayleigh) scattering, photoelectric absorption,
and Compton (inelastic) scattering.
Rayleigh or Coherent scattering occurs when (low energy) X-rays incident on an atom,
thus the atom receives energy only sufficient to cause bound electrons to vibrate at a
frequency similar to that of the incident X-ray photon. The electron re-radiates a photon
of similar energy (frequency), but with small change in direction (angle) without
absorption. (Bailey et al. 2014). This elastic scattering is a source unwanted radiation to
29
diagnostic radiology personnel, but does not exceed 10% of the total interaction processes
in diagnostic radiology. (Davidsson 2010).
Photoelectric absorption occurs when an x-ray photon with sufficient energy (greater than
the binding energy of an orbital electron) incidents on a material, leading to the ionisation
of the material. Ionisation is usually accompanied by the emission of a lower energy
fluorescence, especially when the incident photon has sufficient energy to liberate
electrons closer to the atomic nucleus, as higher orbital electrons drops to fill the holes
created leading to the emission of photons. The probability of photoelectric absorption
increases quite sharply with photon energy just slightly above an electron’s binding
energy due to the following reasons: the increase in the number of electrons available for
interaction, and the occurrence of resonance phenomenon when photon energy just
exceeds the binding energy of a given shell. This phenomenon is known as absorption
edge. (Bailey et al. 2014, Davidsson 2010).
Compton scattering occurs when an incoming photon ejects an atom’s electron (usually
outer electron) and still have sufficient energy to continue, but in a new direction due to
the loss of energy. (Bailey et al. 2014). Two main unwanted effects of Compton scattering
are: image contrast reduction (which should be removed by collimation before detector),
and major radiation source for radiology personnel. (Davidsson 2010).
When an X-ray photon Io, incident on a matter, the above absorption processes contribute
to the total attenuation of the X-ray flux passing through the matter to produce a new
photon I, which is related by the equation 1 below (Davidsson 2010):
I = Io e - µt (1)
where, Io is the incident photon, I is the attenuated photon, t (cm) is the material thickness,
and µ (cm-1) is the linear attenuation coefficient.
The linear attenuation coefficient µ decreases with increasing incident photon energy in
the diagnostic energy range, increases with matter (tissue) density due to the relative
increase in number of atoms per unit volume, increases with atomic number due to the
increase in nuclear forces for higher proton atoms, and increases sharply for energies just
above the absorption edge energy.
30
Linear attenuation coefficient may be expressed as mass attenuation coefficient by
representing the thickness as mass per unit area by multiplying the thickness by density.
2.6 Hip Geometry
A variety of hip geometrical measurements have shown their relation to hip OA and OP.
Anteroposterior pelvic radiography (Castaño-Betancourt et al. 2013), with patient lying
in the supine position remains the favourite position in most pelvic radiographic studies
(Arokoski et al. 2002, Ipach et al. 2013, Javaid et al. 2009, Merle et al. 2014, Pulkkinen
et al. 2004, Pulkkinen et al. 2010, Thevenot et al. 2014).
In OA studies, upper femur geometrical parameters found to increase the risk of hip OA
are, wider femoral neck diameter (ND) (Arokoski et al. 2002, Bendaya et al. 2015,
Castaño-Betancourt et al 2013, Javaid et al. 2009), a more medial centroid femur position
(Javaid et al. 2009), greater cross-sectional moment of inertia of femur (Javaid et al.
2009), greater section modulus (Javaid et al. 2009), higher sacral slope (Bendaya et al.
2015), higher acetabular angle of Idelberger and Frank (Bendaya et al. 2015), higher
femoral mechanical angle (Bendaya et al. 2015), pistol grip deformity (higher femoral
head eccentricity) (Bendaya et al. 2015, Doherty et al. 2008), right–left asymmetries in
centre-edge acetabular angle (Bendaya et al. 2015), longer hip axis length (HAL)
(Castaño-Betancourt et al 2013), higher pelvic width (Castaño-Betancourt et al 2013),
higher triangular index (femur head asphericity) (Castaño-Betancourt et al 2013),
acetabular dysplasia (lower Wiberg angle or lower acetabular depth (d)) (Castaño-
Betancourt et al 2013, Murphy et al. 1995, Reijman et al. 2005), and abnormal neck shaft
angle (NSA) (extremely higher or lower) (Doherty et al. 2008).
For OP fractures, apart from the generally accepted predisposing factor—lower BMD—
and other complementary factors outlined in the FRAX assessment tool, longer (HAL)
(Duboeuf et al. 1997(cervical), Faulkner et al. 1993 (all), Gnudi et al. 2002 (cervical),
Peacock et al. 1995 (general)), shorter ND (Duboeuf et al. 1997 (trochanteric)), lower
medial calcar femoral cortex width (Partanen et al. 2001 (general), Pulkkinen et al. 2004
(all)), lower femoral shaft cortex width (FSC) (Glüer et al. 1994 (general), Partanen et al.
2001 (general and cervical), Pulkkinen et al. 2004 (all), Pulkkinen et al. 2010 (all)), lower
femoral neck cortex width (FNC) (Glüer et al. 1994 (general), Partanen et al. 2001
31
(general and cervical), Pulkkinen et al. 2004 (all), Pulkkinen et al. 2010 (all)), lower
femoral shaft diameter (FSD) (Partanen et al. 2001 (general and cervical), Pulkkinen et
al. 2004 (general and cervical)), lower trochanteric width (Partanen et al. 2001 (general)),
higher NSA (Alonso et al. 2000 (general), Gnudi et al. 2002 (cervical), Partanen et al.
2001 (general and trochanteric), Pulkkinen et al. 2004 (general and cervical), Pulkkinen
et al. 2010 (general and cervical)), lower outer pelvic diameter (general and trochanteric)
(Partanen et al. 2001), lower inner pelvic diameter (Partanen et al. 2001 (general)), and
higher acetabular width (w) (Partanen et al. 2001 (cervical), Pulkkinen et al. 2004
(cervical)) may also contribute to upper femur fractures.
In this study, the relevance of the following upper femur geometries: NSA, femoral neck
axis length (FNAL), HAL, HD, ND, FNC, hip joint space width (JSW), w, d, acetabular
depth-to-width ratio (ADR), and obturator foramen width (OF), in relation to the
development of hip OA and fracture were analysed.
2.7 Trabecular Texture Analysis
With observed trabecular microarchitectural changes in OA (Buckwalter et al. 2005, Lane
2007, Li et al. 2013) and OP (Cosman et al. 2014, Hernlund et al. 2013, Kanis et al. 2013,
WHO 843 1994) conditions, bone trabecular texture analysis has also become a new
research area for improved diagnosis of OA and OP. In OA, subchondral bone sclerosis
has been reported to shield the deeper trabecular from loading stress, leading to localized
osteoporosis. The net effect of this stress shielding is to cause fine to medium vertical
trabeculae to thin and may be lost, whereas principal load bearing vertical trabecular
thickens. Thickening of the vertical trabecular results in the recording of higher BMD of
trabecular of OA patients. (Papaloucas et al. 2005). The main microarchitectural change
that occur in OP is the aggravated loss of trabecular plates, thus thinning of trabecular
rods. This lead to disordered and fragile skeletal microarchitecture (Cosman et al. 2014).
Fractal-based analysis (Buckland-Wright et al. 1996, Kraus et al. 2009, Kraus et al. 2013,
Messent et al. 2005a, Messent et al. 2005b, Messent et al. 2005c, Messent et al. 2006,
Messent et al. 2007, Papaloucas et al. 2005, Podsiadlo et al. 2008, Wolski et al. 2010)
remains the most common method used in OA radiograph analysis. Signature
Dissimilarity Measure (SDM) (Woloszynski et al. 2012) is also used. In a recent study by
32
Hirvasniemi et al. (2014) the use of Laplacian and LBP was used to distinguish OA
patients from controls. Majority of the texture analysis methods for OA have been applied
on plain knee radiographs.
A variety of texture operators have been used to discriminate OP fractures from controls.
Both in vivo (Benhamou et al. 2001, Rachidi et al. 2008) and in vitro (Chappard et al.
2010, Le Corroller et al. 2012, Huber et al. 2009, Vokes et al. 2006) studies have shown
the potential of radiographic texture analysis to discriminate OP fractures from controls.
Other radiographic texture analysis is effective in the discrimination of OP fractures from
controls when combined with BMD (Kolta et al. 2012 (in vitro), Lespessailles et al. 2008
(in vivo)). Studies by Wilkie et al. (2008) demonstrated the ability of radiographic texture
analysis to monitor osteolysis over time. Chappard D et al. (2005) and Guggenbuhl et al.
(2006) showed that bone texture analysis reflects the true volumetric structure of bones.
The use of gradient (or Laplacian) based texture operators was used to discriminate hip
OP fractures from controls by Pulkkinen et al. (2011) and Thevenot et al. (2014).
In this study, we focus on the use of entropy of LBP and Laplacian based, and
Homogeneity Index of Laplacian based radiographic images of upper femur head and
neck trabeculae for the analyses of the dependence hip OA and hip fracture on trabecular
texture.
33
3. Aims of the Study
The main aims of this study were to investigate:
1. The association of different geometrical features in upper femur measured from
pelvic radiographs to the hip OA and future hip fracture,
2. The relationship between the occurrence of hip OA and hip fractures, and
3. The association of upper femur bone texture parameters to the occurrence of OA and
future hip fracture.
34
4. Materials and Methods
4.1 Study Subjects
The study sample consisted of 125 postmenopausal women, born between 1924 and 1927,
inclusively, who were originally recruited from a larger population-based cohort study
started in 1997 (Korpelainen et al. 2003). The selection was made from 618
postmenopausal who attended a follow-up study in 2006, during which their pelvic
radiograph and BMD measurement of the femoral neck were taken. Participants’ weight,
height, and body mass index (BMI) were also taken. Overexposed radiographs were
excluded from the research. Detailed exclusion criteria can be found from Thevenot et al.
2014.
The fracture history of these selected women between December 1997 and June 2012 was
obtained from hospital discharge registers. By June 2012, 23 of these women had
experienced hip fracture.
The procedures of the study were in accordance with the Helsinki Declaration, and was
approved by the formal ethics committee. Informed and written consent were obtained
from all participants.
4.2 Imaging, OA grading, and measurement of geometrical parameters
Digital anteroposterior radiographic images of participants were acquired using an X-ray
system (DR 9000; Eastman Kodak Company, Rochester, NY) at 70 kV; automatic
exposure; resolution, 2560 x 3072 (pixel size; 0.139 x 0.139 mm2); and 14 bits. Images
were taken with patients in standard supine position and the beam focused on the femoral
head. Radiographs were acquired from the participants right hip (n = 123), except for
patients (n = 2) in which it was not possible to image the right hip, mostly due to the
presence of prosthesis in the right femur.
Public domain software developed by the National Institutes of Health (Image J, version
1.48v; http://rsb.info.nih.gov/ij/) was used to assess geometric parameters: NSA, FNAL,
HD, ND, FNC, HAL, JSW, d, w, ADR, and OF (Figures 6 and 7). The dimensions were
35
determined by using a calibration scale, which consisted of a block of known distance,
included in the radiographs.
Figure 6. Hip geometrical parameters measured from radiographs. A-B, femoral neck
axis length (FNAL); A-F, hip axis length (HAL); B-E, joint space width (JSW); D-DD,
head diameter (HD); C-CC, neck diameter (ND); femoral neck shaft angle (NSA); and
femoral neck cortex thickness (FNC).
36
Figure 7. Hip geometrical parameters measured from radiographs. Acetabulum width (w)
and depth (d); and ‘a’, obturator foramen width (OF).
To evaluate the reproducibility of the geometrical measurements, both intra- and
interoperator of geometrical measurements were done. Two intraoperator measurements
were made at an interval of 4 months for all the geometric parameters. A one-time
measurement of the following geometric parameters: NSA, FNAL, HD, ND and FNC,
for 93 of the images selected randomly by another operator was used to evaluate the
interoperator reproducibility. Both intraoperator measurements were compared with that
of the interoperator values, thus 2 interoperator reproducibility parameters were obtained.
In all measurements, operators were blinded to the fracture, and KL grading of subjects.
BMD measurements were made by using a DXA device (Delphi QDR series; Hologic,
Bedford, Mass) from a standard supine anteroposterior position. The BMD was assessed
at the femoral neck by using the standard protocol of the manufacturer’s software.
OA classification was based on the KL grading (Table 1) (Kellgren and Lawrence 1957).
37
4.3 Evaluation of the Textural Parameters
Processing of acquired radiographic images were performed with a custom-made
algorithm (MATLAB version R2014b, The MathWorks, Inc., Natick, MA, USA) with a
graphical interface that allows the user to rotate the image and select the region of interest
(ROI) on the radiograph.
The ROI (size: 100 x 100 pixels) for the femur neck was at the femur inferomedial side
(neck-to-head transition area) containing the principal compressive fibre group. Fibres
orientation were parallel to the side of the rectangle used in selecting the ROI (Figure 8).
The head ROI (100 x 100 pixels) was at the meeting area of the principal compressive
fibre and principal tensile fibre groups (Figure 8).
The reproducibility of the texture parameters has already been ascertained (Hirvasniemi
et al. 2014, Thevenot et al 2014), with values less than 1.6%.
From the ROIs, Local Binary Pattern (LBP) (Ojala et al. 1996) and second order partial
derivatives (Laplacian (Lap)) were calculated to obtain LBP- and Laplacian-based
images. The Laplacian method has been described thoroughly previously (Thevenot et al.
2014). Briefly, the image was pre-processed first with median filtering (3 x 3) to reduce
image noise and subsequently, disk-shaped morphologic top- and bottom-hat operations
were performed. A Laplacian-based matrix was derived from the ROI and was then
calculated perpendicular to the trabecular main orientation. The values were converted to
a scale between 0 and 1. The original ROI was then multiplied by the square root of the
Laplacian matrix. To obtain the final image, the grey scale values were expanded to the
full dynamic range. In the LBP method, the eight neighbour pixels for each pixel in the
ROI were examined and an 8-bit LBP-value was calculated (Hirvasniemi et al. 2014).
Image entropy (Ent), which is a statistical textural measure of randomness was calculated
for both LBP (Ent_LBP) (Hirvasniemi et al. 2014) and Laplacian (Ent_Lap) (Thevenot
et al. 2014) results from each ROI (neck and head). The entropy for both LBP and
Laplacian results was calculated using the equation (2):
𝐸𝑛𝑡 = − ∑ 𝑃𝑖 log2 𝑃𝑖𝑖 (2)
where Pi is the number of occurrence of the gray-level value i in the image.
38
Figure 8. Regions of interest (ROIs) for the femoral neck and head.
Entropy describes the distribution of the local patterns in the LBP method and the
distribution of the grayscale values in the Laplacian method. If Ent_LBP = 0, there is only
single pattern occurring in the original image, whereas Ent_Lap = 0 means that all pixel
values in the Laplacian-based image are the same.
Another texture measure called Homogeneity Index (HI), was calculated from the
Laplacian based images from each ROI. HI was derived from the gray-level co-
occurrence matrix (GLCM) (Haralick et al. 1973) that was calculated in horizontal
direction (0 degrees) with a distance of one pixel. If all adjacent pixel values in an image
are the same, HI is one.
39
4.4 Statistical Analysis
Statistical analyses were performed using the statistical software SPSS (SPSS version
20.0; SPSS Inc., Chicago, USA). Independent samples t-test was used for the assessment
of differences between groups. The intra- and interoperator reproducibility of the
geometric measurements were performed using the root-mean-square coefficient of
variation according to the following equation (3):
CVRMS = √∑ (CVj)2
nj=1
𝑛 (3)
where, CVj is the individual coefficient of variation for the subject j and n is the number
of images analysed for each (geometric) parameter.
40
5. Results
Table 2 shows the results of the intra- and interoperator reproducibility for the geometrical
parameters evaluated using CVRMS (equation 3). Interoperator HD measurements
produced the best reproducibility results (CVRMS of 0.91%), while intraoperator JSW was
the worst to reproduce (CVRMS of 13.96%).
Table 2. Intraoperator and Interoperator Root Mean Squared Coefficient of Variation
(CVRMS).
From Table 3, it can be seen that the distribution of KL grading in the women with and
without hip fracture was rather similar.
Table 3. Fracture Distribution for the various KL grades.
Fracture
Type
KL0 (Controls) KL1 KL2 KL3 Total
No Fracture
56 (54.9 %) 20 (19.61 %) 21 (20.59 %) 5 (4.9 %) 102
Fracture
(all)
13 (56.52%) 3 (13.04%) 5 (21.74%) 2 (8.7%) 23
Cervical
10 (66.67 %) 2 (13.33 %) 2 (13.33 %) 1 (6.67 %) 15
Trochanteric
3 (37.5 %) 1 (12.5 %) 3 (37.5 %) 1 (12.5%) 8
Total 69 23 26 7 125
Quantity Intraoperator CVRMS
[%]
Interoperator CVRMS1
[%]
Interoperator CVRMS2
[%]
NSA 1.95 0.95 2.29
FNAL 1.30 1.15 2.10
HD 1.14 0.91 1.31
ND 2.29 4.08 2.84
FNC 5.42 11.63 12.13
HAL 1.73
JSW 13.96
OF 1.84
w 2.64
d 4.98
41
The differences between OA and control groups by the various geometrical parameters,
weight, height, BMI, and neck BMD are presented in Table 4. KL1 grade group was
eliminated from the analyses to boost the confidence of the OA and control groupings.
The ND and JSW had the most significant difference between controls and OA. The
differences in weight, BMI, neckBMD and FNC between OA and controls were also
statistically significant (p < 0.05).
Table 4. Means of OA and Controls for the various geometrical parameters, weight,
height, BMI, and neck BMD.
Table 5 shows the values for various texture parameters for OA and control groups,
without KL1 grades. Only clear images were considered for this analysis. Neck HI_Lap
Parameter
Controls (n = 69) OA (n = 33) p-value
Weight (kg) 64.95
70.09
0.041
Height (cm) 156.36
156.2
0.927
BMI (kg/m2) 26.51
28.635
0.017
neckBMD (g/cm2) 0.61
0.676
0.005
NSA (Deg) 127.2
126.7
0.615
FNAL (mm) 99.06 99.44
0.767
HD (mm) 46.94
47.73
0.146
ND (mm) 34.14
36.22
<0.001
FNC (mm) 4.97
5.391
0.038
JSW (mm) 5.84
4.166
<0.001
HAL (mm) 114.9
115.9
0.535
ADR 0.330
0.324
0.500
OF (mm) 31.73
32.30
0.781
w (mm) 55.53
55.92
0.607
d (mm) 18.31 18.03 0.603
42
produced the highest relative difference between the means of OA and controls (p =
0.001). All other texture parameters, except neck Ent_Lap, differed significantly between
the groups (p < 0.05).
Table 5. Means of OA and Controls for the various texture parameters (clear images
only).
Means of the various geometrical features, BMI, neckBMD, weight and height for
fracture (n = 23) and control (non-fracture) subjects are presented in Table 6. Acetabulum
width (w) of patients recorded the relative highest difference between fracture and control
subjects (p = 0.011). Femur neck BMD, FNAL, JSW, and HAL also showed a significant
difference between fracture and controls subjects (p < 0.05). When only cervical fractures
were considered (Table 7), only the differences in JSW and w were statistically significant
(p < 0.05). Similarly, with only trochanteric fractures, the differences in neckBMD, HD,
ND and HAL were statistically significant (p < 0.05) (Table 8).
Parameter
Controls (n = 68) OA (n = 30) p-value
Ent_LBP_neck
6.845
6.829
0.477
Ent_Lap_neck
6.558
6.211
0.003
HI_Lap_neck
0.781
0.813
0.001
Ent_LBP_head
6.809
6.743
0.030
Ent_Lap_head
5.873
5.542
0.002
HI_Lap_head 0.836 0.863 0.005
43
Table 6. Geometrical parameters, weight, BMI and neckBMD for fracture and control
subjects.
Parameter
Controls (n = 102) Fracture (n = 23) p-value
Weight (kg) 66.63
68.72
0.468
Height (cm) 155.6
157.9
0.090
BMI (kg/m2) 27.50
27.38
0.910
neckBMD (g/cm2) 0.65
0.594
0.037
NSA (Deg) 127.0
127.5
0.678
FNAL (mm) 98.44
101.1
0.046
HD (mm) 46.95
48.04
0.060
ND (mm) 34.58
35.65
0.104
FNC (mm) 5.086
4.89
0.401
JSW (mm) 5.15
6.15
0.017
HAL (mm) 114.2
117.7
0.046
ADR 0.327
0.321
0.582
OF (mm) 31.93
29.16
0.196
w (mm) 55.21
57.24
0.011
d (mm) 18.02 18.39 0.540
44
Table 7. Geometrical parameters, weight, BMI and neckBMD for cervical fracture and
control subjects.
Parameter Controls (n = 102) Cervical Fracture
(n = 15)
p-value
Weight (kg) 66.63
69.81
0.363
Height (cm) 155.6
159.4
0.070
BMI (kg/m2) 27.50
27.31 0.884
neckBMD (g/cm2) 0.648
0.620
0.370
NSA (Deg) 127.0
127.9
0.533
FNAL (mm) 98.44
100.6
0.167
HD (mm) 46.95
47.63
0.326
ND (mm) 34.58
34.81
0.766
FNC (mm) 5.086
4.958
0.649
JSW (mm) 5.147
6.247
0.029
HAL (mm) 114.2
116.4
0.271
ADR 0.327
0.327
0.992
OF (mm) 31.93
28.81
0.236
w (mm) 55.21
57.40
0.024
d (mm) 18.02 18.78 0.307
45
Table 8. Geometrical parameters, weight, BMI and neckBMD for trochanteric fracture
and control subjects.
The means of the various texture parameters for cervical fractures and controls (without
trochanteric) fractures is presented in Table 9, along with their p-values. The statistically
most significant differences in texture parameters were in Ent_LBP_head and
HI_Lap_head with p-values of 0.001 and 0.029 respectively.
Parameter Controls (n =102) Trochanteric
Fracture (n = 8)
p-value
Weight (kg) 66.63
66.68
0.992
Height (cm) 155.6
155.2
0.852
BMI (kg/m2) 27.50
27.51
0.994
neckBMD (g/cm2) 0.648
0.544
0.011
NSA (Deg) 127.0
126.8
0.891
FNAL (mm) 98.44
102.1
0.082
HD (mm) 46.95
48.80
0.036
ND (mm) 34.58
37.23
0.012
FNC (mm) 5.086
4.763
0.380
JSW (mm) 5.147
5.956
0.225
HAL (mm) 114.2
120.0
0.031
ADR 0.327
0.311
0.333
OF (mm) 31.93
29.81
0.556
w (mm) 55.21
56.93
0.174
d (mm) 18.02 17.67 0.706
46
Table 9. Texture parameters for cervical fractures and controls (without trochanteric
fractures) (clear images only).
Parameter Control (n = 99) Cervical Fracture
(n = 13)
p-value
Ent_LBP_neck 6.838
6.793
0.162
Ent_Lap_neck 6.445
6.279
0.301
HI_Lap_neck 0.789
0.808
0.155
Ent_LBP_head 6.804
6.667
0.001
Ent_Lap_head 5.805
5.527
0.059
HI_Lap_head 0.839 0.867 0.029
When comparing texture parameters between trochanteric fractures and controls, no
significant differences were found.
47
6. Discussion
In this study, we investigated the association of different geometrical features in upper
femur measured from pelvic radiographs to the hip OA and future hip fracture. Also, we
investigated the relationship between the occurrence of hip OA and hip fractures, and
association of upper femur bone texture parameters to the occurrence of OA and future
hip fracture. Based on previous studies on geometrical features in upper femur, it was
hypothesized that lower NSA (Moore al. 1994) and longer HAL (Castaño-Betancourt et
al. 2013) predisposes a person to OA, whereas higher NSA and lower cortical thickness
(Partanen et al. 2001, Pulkkinen et al. 2004, Pulkkinen et al. 2010, Pulkkinen et al. 2011)
predisposes a person to increased risk of cervical fracture. One main factor that
encouraged bone texture analysis was the changes that occur in bone structure in
osteoporotic (Cosman et al. 2014, Hernlund et al. 2013, Kanis et al. 2013) and
osteoarthritic (Buckwalter et al 2005, Lane 2007, Egloff et al. 2012) conditions.
OA patients (KL ≥2) had less hip fractures, but due to the limited sample size in subgroup
analyses, the current study does not give confirmation whether persons having OA have
reduced risk to develop hip fractures. Blain et al. (2008) found a substantial increase in
trabecular bone volume of postmenopausal OA (compared to OP) patients, and that OA
patients generally have thicker cortical thickness, all of which decreases the tendency of
a bone to fracture. This inverse relationship is in agreement with some previous studies
(Cumming and Klineberg 1993, Franklin et al. 2010, Vestergaard et al. 2009), but
contrary to those from other studies made (Arden et al. 1999 and Jones et al. 1995).
The current results indicate that weightier women are at a higher risk of developing OA
compared to women of low weight. Blain et al. (2008) and Lohmander et al. (2009) also
found higher prevalence of hip OA among overweight people. OA data on other body
parts like wrist and knee also found persons with higher weight to be at risk of developing
OA (Carman et al. 1994, Felson et al. 1997, Spector et al. 1994), and that obesity even
accelerates the progression of OA (Cooper et al. 2000, Schouten et al. 1992).
Biomechanically, the reason for this increase in OA among overweight persons is due to
the high force exercised on the joint, making the articular cartilage and subchondral bone
more susceptible to damage. Rising glucose and insulin levels due to excess body fat are
48
also attributed to joint inflammation, thereby increasing the progression of OA (Arthritis
Research Campaign).
Height of participants did not show any dependence on OA. Liu et al. (2007) found the
probability of having a hip replacement, which OA is a considered a major precursor to
it, to increase among taller women. According to Hunter et al. (2005) (and Blain et al.
2008 in comparison with OP patients) tall people are at risk of developing knee OA.
Biomechanically, this independence relation is expected, as tall people usually have
larger bones and thick cartilage, which experience high impact forces. Short people have
small bones and thin cartilage, which experience low impact forces. Conversely,
extremely tall and short people may be at risk of developing OA.
From our study elderly women with higher BMI were found to be at higher risk of having
OA. A study by Jiang et al. (2011) and Oliveria et al. (1999) also produced similar results.
Results for OA dependence on BMI is relatively tied to obesity as obese persons have
higher BMI.
In our study, participants with higher femoral neck BMD were found to be at higher risk
of developing OA. Arden and Nevitt (2006), Blain et al. (2008), and Nevitt et al. (1995)
found the same trend. Even though in our study OA subjects were found to have higher
FNC too, which may contribute to higher BMD value with areal BMD measurement,
observed thickening of larger weight bearing trabeculae could also result to higher BMD
among OA subjects (Papaloucas et al. 2005).
In our research, NSA of controls was larger, but was not significantly different compared
to OA subjects, which was in agreement with that obtained by Reikerås and Høiseth
(1982) and Bendaya et al. (2015). The HD of OA patients in our group were slightly
higher than controls (p = 0.146), similar to results obtained by Bendaya et al. (2015) and
Jonhson et al. (2012).
ND of OA patients were significantly higher than in control group (p < 0.005). Arokoski
et al. (2002) and Johnson et al. (2012) also found higher femur ND among OA patients.
Thus ND could be a diagnostic factor in predicting OA among patients. The reason for
this has not been scientifically ascertained, but biomechanically one may speculate that
since OA patients are generally weightier, bone remodelling contribute to the increase in
ND to accommodate the higher impact force from the body’s upper parts.
49
There was a significant positive correlation between FNC and OA. Blain et al. (2008) and
Turmezei et al. (2013) found a similar trend, but Preidler et al. (1997) found no correlation
between cortical thickness and OA. The higher FNC among OA subjects could be the
result of the higher weight and BMI among OA patients, thus triggering more osteoblast
activity, leading to higher periosteal appositional deposition. Joint sites of higher cortical
content can exert higher force on joint articulating structures, which may eventually lead
to damage of articular cartilage, thus accelerating the progression of OA.
JSW for OA subjects were significantly lower than controls. This result was expected
since in the KL grading JSW is one of the measures that affect the KL grade. Joint space
narrowing remains one of the main predictors of OA. HAL of OA subjects was similar to
controls, which was contrary to the findings by Castaño-Betancourt et al. (2013).
In this study, Ent_Lap_Neck was significantly higher for controls than for OA subjects.
The cause of this higher value is that, the normal bone texture is intuitively more random
than OA bones as bone fibrils are laid in alternate fashion (Eriksen et al.1994, Seelay et
al. 1998). Higher values of Ent_LBP_head and Ent_Lap_head could be caused by the
same randomness among normal bone trabecular texture, but the veracity of femur head
entropy values need further verification from separate femur head studies with no
acetabulum influence.
The higher homogeneity index HI_Lap_neck could be attributed to increase in load
bearing (vertical) trabeculae thickness and cross-connectivity among OA patients,
especially in advanced OA (Papaloucas et al. 2005, Buckland-Wright et al. 1996). For
higher HI_Lap_head among OA, fractal signature analysis, which measures horizontal
and vertical trabeculae separately (Messent et al. 2005a), of the same ROI by Papaloucas
et al. 2005 showed increase in vertical trabeculae thickness and cross-connectivity, thus
higher homogeneity among OA femur head. The increased homogeneity index of OA
subjects may also be further aggravated by subchondral bone sclerosis which contribute
to increase in BMD of OA patients, thus the bone texture becomes more homogenous.
But due to the overlap of the acetabulum for the femoral head ROI, further studies may
be needed to ascertain the higher value of HI_Lap_head.
In general, subjects with lower femoral neck BMD were more prone to fracture. Also,
FNAL, JSW, HAL, and w were higher among subjects with fractures. Lower BMD
(osteoporosis) is regarded as one of the high risk factors of a fracture, supported by
50
numerous studies (Cosman et al. (2014), Kanis et al. (2013), Marshall et al. (1996)) due
to the effect it has on bone structure, by making it porous and fragile. Fracture grouping
elucidated that femoral neck BMD is a predictor of trochanteric fractures compared to
cervical fractures. This shows one of the limitations of (projectional) DXA, by
overestimating the neck BMD due to the cortical thickness of the femur cervix (for there
were more cervical fractures even with least difference in average BMD) and also site
specific BMD as the best indicator of fracture (Pulkkinen et al. 2004). FNAL of general
hip fractured cases were significantly higher than non-fractured cases.
NSA was not a general predictor of fracture in our study. Compared with other studies
where NSA was a predictor of hip fractures (Alonso et al. 2000, Gnudi et al. 2002,
Partanen et al. 2001, Pulkkinen et al. 2004, Pulkkinen et al. 2010, Thevenot et al. 2014),
our study population generally had lower NSA than in the other studies.
FNAL was a significant predictor of hip fractures in general, with longer length being at
those at higher risk. However, when cervical and trochanteric fractures were considered
separately, the significance disappeared. In studies by Pulkkinen et al. (2004) and
Pulkkinen et al (2010) no statistical significance was found FNAL in relation to fractures,
but controls had slightly higher FNAL.
HD in trochanteric fracture cases was higher than in the non-fractured hips. Partanen et
al. (2001), and Pulkkinen et al. (2004) found no difference between HD of controls and
fractured cases. In a comparison of HD of cervical and trochanteric fractures, Patton et al
(2006) found no significant difference between the HDs of cervical and trochanteric
fractured patients. Calis et al. (2004) found no difference between the HDs of controls
and fractured subjects.
ND was statistically significantly higher only when trochanteric fractures were
considered. Pulkkinen et al. (2004) also obtained similar results, with higher ND for
trochanteric fractures, but was not statistically significant. Calis et al. (2004) found ND
of fractured patients to be significantly higher than controls.
JSW was higher and statistically significant for all fractures and for cervical fractured
subjects than for controls, but not on trochanteric fractures.
HAL was longer for all fractures and for trochanteric fractures as well, but not for cervical
fractures. Gnudi et al. (2002) found a positive correlation between longer HAL and
incidence of cervical fractures. Pulkkinen et al. (2004), Pulkkinen et al. (2010), and
51
Pulkkinen et al. (2011) found no significant correlation between HAL and fractures, but
HAL of cervical fractured patients were generally longer. Longer HAL and higher ND in
trochanteric fracture group may due to the moment exerted on the femur shaft at the
trochanteric side, since the higher ND (even with lower BMD) is able to shield the femur
neck from the forces from the upper extremity, thus making the trochanteric more
susceptible to fractures in such cases.
Our study did not find ADR, d, or OF to be a predictor of a hip fracture. The parameter
w for fractured subjects were higher than for controls, and was statistically significant for
fractures in general and for cervical fractures, but not for trochanteric fractures when
compared with controls. Partanen et al. 2001, and Pulkkinen et al. 2004 also found higher
w to be a predictor cervical fractures. Thus more studies may be done to ascertain the
validity of our findings.
The Ent_LBP_head and HI_Lap_head were significant predictors for cervical fractures,
with lower and higher values for cervical fractures, respectively. This inverse relationship
may not be considered as a research finding, due to the overlapping of the acetabulum
and the femur head at the site of texture analysis.
With the aging population on the increase, accurate diagnosis of the diseases of the aged
are also needed, and with this method in which in vivo samples were used, it could be a
welcoming research results of the health care system of many countries. Incorporating it
into diagnostic radiographic machines will be the optimum choice to relief the cost of
obtaining and analysing of samples.
Our study has number of limitations. Since all images were taken in 2D, geometrical
parameters were dependent on hip positioning. More detailed analysis on the relation of
these geometrical parameters to OA and OP should be studied in 3D environment in
future. Texture analysis from the femoral head is problematic from projectional images.
This is because the acetabulum is summed in the image and the femur bone structure is
not distinct there. Therefore the texture measures from the head are prone to errors. Also,
the soft (fat) tissue might have produced errors in the image. Furthermore, the manual
selection of ROIs and the inability to monitor the progress of OA are other limitations of
this research. The results cannot be generalized into whole population since relatively
elderly women were studied, and the number of OA and fracture cases were low. The
52
population used this study might have been biased since all fracture cases were selected
and were presented in OA analyses.
53
7. Summary
In this in vivo study we found higher ND and FNC, and lower JSW among OA subjects.
Higher weight, BMI, and femur neck BMD were also associated with OA. Subjects with
general hip fractures had lower neck BMD, and higher FNAL, JSW, HAL, and w. For
subjects with cervical fractures, higher JSW and w were found among them. Lower neck
BMD and HD, and higher ND and HAL were found among subjects with trochanteric
fractures.
Our study did not ascertain whether OA is predisposing factor for hip fractures or not,
mainly due to the limited sample population used.
Lower Ent_Lap_Neck and higher HI_Lap_neck were found among OA subjects. Lower
Ent_LBP_head and Ent_Lap_head, and higher HI_Lap_head found among OA subjects,
may have been due to the overlap of the acetabulum in the ROI for the head. Similarly,
for subjects with cervical fractures, Ent_LBP_head was lower, and HI_Lap_head was
higher, but this inverse relationship may also have been due to the overlap of the
acetabulum in the ROI of the femur head.
Our research shows that upper femur geometry may play a role in the initiation and
progression of OA and OP. The results support the use of bone fibers of the main
compressive system in the prediction of OA. The high discrimination of the method,
coupled with its high reproducibility make it a robust and cheap method that might be
utilised in clinical diagnosis. Another good feature of the bone texture analysis method is
that it is less affected by variation in imaging parameters (like X-ray tube current, image
exposure time, grid parameters) than direct evaluation of grayscale values.
Further studies with a larger study group is needed to ascertain the clinical validity of the
method. Other strong indicators of OA might be combined with it to improve the
specificity of the method.
54
References
1. A level Medical Physics - X rays- the X ray tube, 2013 (Video file), Accessed
from <https://www.youtube.com/watch?v=QqJqKwm3Vd0> (Accessed 24 July
2015).
2. Alonso CG, Curiel MD, Carranza FH, Cano RP, Perez AD, and the Multicenter
Project for Research in Osteoporosis (2000). Femoral bone mineral density, neck-
shaft angle and mean femoral neck width as predictors of hip fracture in men and
women. Osteoporos Int. 2000; 11(8): 714-20.
3. Altman R, Alarcón G, Appelrouth D, Bloch D, Borenstein D, Brandt K, Brown C,
Cooke TD, Daniel W, Feldman D, et al. (1991). The American College of
Rheumatology criteria for the classification and reporting of osteoarthritis of the
hip. Arthritis Rheum. 1991 May; 34 (5):505-14.
4. Arden N, Nevitt MC (2006). Osteoarthritis: epidemiology. Best Pract Res Clin
Rheumatol. 2006 Feb; 20 (1):3-25.
5. Arden NK, Nevitt MC, Lane NE, Gore LR, Hochberg MC, Scott JC, Pressman
AR, Cummings SR (1999) .Osteoarthritis and risk of falls, rates of bone loss, and
osteoporotic fractures. Study of Osteoporotic Fractures Research Group. Arthritis
Rheum. 1999 Jul; 42(7):1378-85.
6. Arokoski JP, Arokoski MH, Jurvelin JS, Helminen HJ, Niemitukia LH, Kröger H
(2002). Increased bone mineral content and bone size in the femoral neck of men
with hip osteoarthritis. Ann Rheum Dis. 2002 Feb; 61 (2):145-50.
7. Arthritis Research Campaign. ISBN 978-1-901815-14-6. Osteoarthritis and
Obesity (Accessed 5 August 2015).
55
8. Bailey DL, Humm JL, Todd-Pokropek, van Aswegen A (2014). Nuclear Medicine
Physics: A Handbook for Teachers and Students, IAEA, 2014, p 1-48.
9. Bauer JS, Kohlmann S, Eckstein F, Müller D, Lochmüller EM, Link TM (2006).
Structural analysis of trabecular bone of the proximal femur using multislice
computed tomography: a comparison with dual X-ray absorptiometry for
predicting biomechanical strength in vitro. Calcif Tissue Int. 2006 Feb; 78(2):78–
89.
10. Baum T, Carballido-Gamio J, Huber MB, Müller D, Monetti R, Räth C, Eckstein
F, et al (2010). Automated 3D trabecular bone structure analysis of the proximal
femur—prediction of biomechanical strength by CT and DXA. Osteoporos Int.
2010 Sep; 21(9): 1553–1564.
11. Beiser A (2003a) Concepts of modern Physics 6ed McGraw-Hill, New York, pp
52-91.
12. Beiser A (2003b). Concepts of modern Physics 6ed McGraw-Hill, New York, pp
228-265.
13. Bendaya S, Lazennec JY, Anglin C, Allena R, Sellam N, Thoumie P, Skalli W
(2015). Healthy vs. osteoarthritic hips: a comparison of hip, pelvis and femoral
parameters and relationships using the EOS® system. Clin Biomech (Bristol,
Avon). 2015 Feb; 30 (2):195-204.
14. Benhamou CL, Poupon S, Lespessailles E, Loiseau S, Jennane R, Siroux V, Ohley
W, Pothuaud L (2001). Fractal analysis of radiographic trabecular bone texture
and bone mineral density: two complementary parameters related to osteoporotic
fractures. J Bone Miner Res. 2001 Apr; 16 (4):697-704.
15. Blain H, Chavassieux P, Portero-Muzy N, Bonnel F, Canovas F, Chammas M,
Maury P, Delmas PD (2008). Cortical and trabecular bone distribution in the
femoral neck in osteoporosis and osteoarthritis. Bone. 2008 Nov; 43(5):862-8.
56
16. Blake GM, Fogelman I (2007) Role of dual-energy X-ray absorptiometry in the
diagnosis and treatment of osteoporosis. J Clin Densitom 2007 Jan-Mar; 10
(1):102–110.
17. Boehm HF, Lutz J, Korner M, Mutschler W, Reiser M, Pfeifer KJ (2009). Using
Radon transform of standard radiographs of the hip to differentiate between
postmenopausal women with and without fracture of the proximal femur.
Osteoporos Int. 2009 Feb; 20(2):323–33.
18. Bonewald LF (2006). Mechanosensation and Transduction in Osteocytes.
Bonekey Osteovision. 2006 Oct; 3(10): 7–15.
19. Brodsky B, Persikov AV (2005) Molecular structure of the collagen triple helix.
Adv Protein Chem. 2005; 70:301-39.
20. Buckland-Wright JC, Lynch JA, Macfarlane DG (1996). Fractal signature
analysis measures cancellous bone organisation in macroradiographs of patients
with knee osteoarthritis. Ann Rheum Dis. 1996 Oct; 55 (10):749-55.
21. Buckland-Wright C (1999). Radiographic assessment of osteoarthritis:
comparison between existing methodologies. Osteoarthritis Cartilage. 1999 Jul; 7
(4), 430–3.
22. Buckwalter JA, Mankin HJ (1997). Articular cartilage: tissue design and
chondrocyte-matrix interactions. Instr Course Lect. 1998; 47:477-86.
23. Buckwalter JA, Mankin HJ, Grodzifnsky AJ (2005). Articular cartilage and
osteoarthritis. Instr Course Lect 2005; 54:465-480.
24. Buckwalter JA, Martin JA (2006). Osteoarthritis. Adv Drug Deliv Rev. 2006 May
20; 58 (2) 150–167.
57
25. Burr DB (2002). Targeted and nontargeted remodeling. Bone. 2002 Jan; 30(1):2-
4.
26. Burr DB, Gallant MA (2012). Bone remodelling in osteoarthritis. Nat Rev
Rheumatol 2012, 8:665–673.
27. Burstein D, Gray M (2003). New MRI techniques for imaging cartilage. J Bone
Joint Surg Am. 2003; 85-A Suppl 2: 70-77.
28. Calis HT, Eryavuz M, Calis M (2004). Comparison of femoral geometry among
cases with and without hip fractures. Yonsei Med J. 2004 Oct 31; 45 (5): 901-7.
29. Carman WJ, Sowers M, Hawthorne VM, Weissfeld LA (1994) Obesity as a risk
factor for osteoarthritis of the hand and wrist: a prospective study. Am J
Epidemiol. 1994 Jan 15; 139 (2):119-29.
30. Carmona RH (2004). US Surgeon General’s Report: Bone Health and
Osteoporosis.
31. Castañeda S, Roman-Blas JA, Largo R, Herrero-Beaumont G (2012).
Subchondral bone as a key target for osteoarthritis treatment. Biochem Pharmacol
2012, 83:315–323.
32. Castaño-Betancourt MC, Van Meurs JB, Bierma-Zeinstra S, Rivadeneira F,
Hofman A, Weinans H, Uitterlinden AG, Waarsing JH (2013). The contribution
of hip geometry to the prediction of hip osteoarthritis. Osteoarthritis Cartilage.
2013 Oct; 21 (10):1530-6.
33. Chappard C, Brunet-Imbault B, Lemineur G, Giraudeau B, Basillais A, Harba R,
et al (2005). Anisotropy changes in post-menopausal osteoporosis:
characterization by a new index applied to trabecular bone radiographic images.
Osteoporos Int. 2005 Oct; 16(10): 1193–202.
58
34. Chappard C, Bousson V, Bergot C, Mitton D, Marchadier A, Moser T, Benhamou
CL, Laredo JD (2010). Prediction of femoral fracture load: cross-sectional study
of texture analysis and geometric measurements on plain radiographs versus bone
mineral density. Radiology. 2010 May; 255 (2):536-43.
35. Chappard D, Guggenbuhl P, Legrand E, Baslé MF, Audran M (2005). Texture
analysis of X-ray radiographs is correlated with bone histomorphometry. J Bone
Miner Metab. 2005; 23 (1):24-9.
36. Clark JM, Huber JD (1990). The structure of the human subchondral plate. J Bone
Joint Surg Br. 1990 Sep; 72(5):866-73.
37. Clarke B (2008). Normal bone anatomy and physiology. Clin J Am Soc Nephrol.
2008 Nov; 3 Suppl 3:S131-9.
38. Cooper C, Snow S, McAlindon TE, Kellingray S, Stuart B, Coggon D, Dieppe PA
(2000). Risk factors for the incidence and progression of radiographic knee
osteoarthritis. Arthritis Rheum. 2000 May; 43(5):995-1000.
39. Cosman F, de Beur SJ, LeBoff MS, Lewiecki MS, Tanner B, Randall S, Lindsay
R (2014). Clinician's guide to prevention and treatment of osteoporosis.
Osteoporos Int. 2014 Oct; 25(10):2359-81.
40. Cumming RG, Klineberg RJ (1993). Epidemiological study of the relation
between arthritis of the hip and hip fractures. Ann Rheum Dis. 1993 Oct; 52
(10):707-10.
41. Cummings SR, Black DM, Nevitt MC, Browner W, Cauley J, Ensrud K, Genant
HK, Palermo L, Scott J, Vogt TM (1993). Bone density at various sites for
prediction of hip fractures. The Study of Osteoporotic Fractures Research Group.
Lancet. 1993 Jan 9; 341 (8837):72–5.
59
42. Damen J, Schiphof D, Wolde ST, Cats HA, Bierma-Zeinstra SM, Oei EH (2014).
Inter-observer reliability for radiographic assessment of early osteoarthritis
features: the CHECK (cohort hip and cohort knee) study. Osteoarthritis Cartilage.
2014 Jul; 22(7):969-74.
43. Davidsson L (2010) IAEA human health series No. 15—Dual Energy X Ray
Absorptiometry for Bone Mineral Density and Body Composition Assessment.
44. Doherty M, Courtney P, Doherty S, Jenkins W, Maciewicz RA, Muir K, Zhang
W (2008). Nonspherical femoral head shape (pistol grip deformity), neck shaft
angle, and risk of hip osteoarthritis: a case-control study. Arthritis Rheum. 2008
Oct; 58 (10):3172-82.
45. Duboeuf F, Hans D, Schott AM, Kotzki PO, Favier F, Marcelli C, Meunier PJ,
Delmas PD (1997). Different morphometric and densitometric parameters predict
cervical and trochanteric hip fracture: the EPIDOS Study. J Bone Miner Res. 1997
Nov; 12 (11):1895-902.
46. Duckworth T, Blundell CM (2010). Orthopaedics and Fractures 4ed, Wiley-
Blackwell, West Sussex, UK pages 25-48.
47. Egloff C, Hügle T, Valderrabano V (2012). Biomechanics and pathomechanisms
of osteoarthritis. Swiss Med Wkly. 2012 Jul 19;142: w13583
48. Eriksen EF, Axelrod DW, Melsen F (1994). Bone histomorphometry, New York,
Raven Press, 1994, pp 1 –12.
49. Faulkner KG, Cummings SR, Black D, Palermo L, Glüer CC, Genant HK (1993).
Simple measurement of femoral geometry predicts hip fracture: the study of
osteoporotic fractures. J Bone Miner Res. 1993 Oct; 8 (10):1211-7.
60
50. Felson DT, Zhang Y, Hannan MT, Naimark A, Weissman B, Aliabadi P, Levy D.
(1997) Risk factors for incident radiographic knee osteoarthritis in the elderly: the
Framingham Study. Arthritis Rheum. 1997 Apr; 40(4):728-33.
51. Franklin J, Englund M, Ingvarsson T, Lohmander S (2010). The association
between hip fracture and hip osteoarthritis: A case-control study. BMC
Musculoskelet Disord. 2010 Nov 26; 11:274.
52. Fransen M, McConnell S, Hernandez-Molina G, Reichenbach S (2014). Exercise
for osteoarthritis of the hip. Cochrane Database Syst Rev. 2014 Apr 22;
4:CD007912.
53. Frost A, Pavlou G, Richards PJ, Belcher J, Jasani V (2010). Influence of
Acetabular and Femoral Version on Fractures of the Femoral Neck. Clin Orthop
Relat Res. 2010 Aug; 468(8): 2224–2229.
54. Genant HK, Lang TF, Engelke K, Fuerst T, Glüer C, Majumdar S, Jergas M
(1996). Advances in the noninvasive assessment of bone density, quality, and
structure:Calcif Tissue Int. 1996;59 Suppl 1:S10-5.
55. Giangregorio LM, Leslie WD, Lix LM, Johansson H, Oden A, McCloskey E,
Kanis JA (2012). FRAX underestimates fracture risk in patients with diabetes. J
Bone Miner Res. 2012 Feb; 27 (2):301–8.
56. Glüer CC, Cummings SR, Pressman A, Li J, Glüer K, Faulkner KG, Grampp S,
Genant HK (1994). Prediction of hip fractures from pelvic radiographs: the study
of osteoporotic fractures. The Study of Osteoporotic Fractures Research Group. J
Bone Miner Res. 1994 May; 9 (5):671-7.
57. Gnudi S, Ripamonti C, Lisi L, Fini M, Giardino R, Giavaresi G (2002). Proximal
femur geometry to detect and distinguish femoral neck fractures from trochanteric
fractures in postmenopausal women. Osteoporos Int. 2002 Jan; 13 (1) 69-73.
61
58. Gnudi S, Sitta E, Pignotti E. (2012). Prediction of incident hip fracture by femoral
neck bone mineral density and neck-shaft angle: a 5-year longitudinal study in
post-menopausal females. Br J Radiol. 2012 Aug; 85(1016):e467-73.
59. Goldring M., Goldring S. (2010). Articular cartilage and subchondral bone in the
pathogenesis of osteoarthritis. Ann NY Acad Sci 2010, 1192:230–237.
60. Goldring SR (2012). Alterations in periarticular bone and cross talk between
subchondral bone and articular cartilage in osteoarthritis. Ther Adv Musculoskelet
Dis. 2012 Aug; 4 (4):249-58.
61. Gregory JS, Stewart A, Undrill PE, Reid DM, Aspden RM (2004). Identification
of hip fracture patients from radiographs using Fourier analysis of the trabecular
structure: a cross-sectional study. BMC Med Imaging. 2004 Oct 6; 4(1):4.
62. Guggenbuhl P, Bodic F, Hamel L, Baslé MF, Chappard D (2006). Texture
analysis of X-ray radiographs of iliac bone is correlated with bone micro-CT.
Osteoporos Int. 2006; 17 (3):447-54
63. Günther KP, Sun Y (1999). Reliability of radiographic assessment in hip and knee
osteoarthritis. Osteoarthritis Cartilage. 1999 Mar; 7(2):239-46.
64. Haralick RM, Shanmugam K, Dinstein I. Textural features for image
classification. IEEE Trans Syst Man Cybern 1973; smc 3:610-21.
65. Hendee WR, Ritenour ER (2002a). Medical imaging Physics 4ed, Wiley-Liss Inc,
New York, USA, pp 117-125.
66. Hendee WR, Ritenour ER (2002b). Medical imaging Physics 4ed, Wiley-Liss Inc,
New York, USA, pp 127-141
62
67. Hendee WR, Ritenour ER (2002c). Medical imaging Physics 4ed, Wiley-Liss Inc,
New York, USA, pp 217-234.
68. Herlidou S, Grebe R, Grados F, Leuyer N, Fardellone P, Meyer ME (2004).
Influence of age and osteoporosis on calcaneus trabecular bone structure: a
preliminary in vivo MRI study by quantitative texture analysis. Magn Reson
Imaging. 2004 Feb; 22 (2):237–43.
69. Hernlund E, Svedbom A, Ivergård M, Compston J, Cooper C, Stenmark J,
McCloskey EV, Jöhnsson B, Kanis JA (2013). Osteoporosis in the European
Union: medical management, epidemiology and economic burden. Arch
Osteoporos. 2013; 8:136.
70. Hinman RS, Heywood SE, Day AR (2007). Aquatic physical therapy for hip and
knee osteoarthritis: Results of a single-blind randomized controlled trial. Phys
Ther. 2007 Jan; 87(1):32-43.
71. Hip Fracture 2015, Accessed from <http://www.drugs.com/health-guide/hip-
fracture.html> (Accessed 14 July 2015).
72. Hirvasniemi J, Thevenot J, Immonen V, Liikavainio T, Pulkkinen P, Jämsä T,
Arokoski J, Saarakkala S (2014). Quantification of differences in bone texture
from plain radiographs in knees with and without osteoarthritis. Osteoarthritis
Cartilage. 2014 Oct; 22(10):1724-31.
73. Holmdahl DE., Ingelmark BE. (1950). The contact between the articular cartilage
and the medullary cavities of the bone. Acta Orthop Scand 1950, 20:156–165.
74. Holopainen JT, Brama PA, Halmesmaki E, Harjula T, Tuukkanen J, van Weeren
PR, Helminen HJ, Hyttinen MM (2008) Changes in subchondral bone mineral
density and collagen matrix organization in growing horses. Bone. 2008 Dec; 43
(6):1108-14.
63
75. Huber MB, Carballido-Gamio J, Bauer JS, Baum T, Eckstein F, Lochmüller EM,
Majumdar S, Link TM (2008). Proximal femur specimens: automated 3D
trabecular bone mineral density analysis at multidetector CT--correlation with
biomechanical strength measurement. Radiology. 2008 May; 247(2):472-81.
76. Huber MB, Carballido-Gamio J, Fritscher K, Schubert R, Haenni M, Hengg C,
Majumdar S, Link TM (2009). Development and testing of texture discriminators
for the analysis of trabecular bone in proximal femur radiographs. Med Phys. 2009
Nov; 36 (11):5089-98.
77. Hudelmaier M, Kollstedt A, Lochmüller EM, Kuhn V, Eckstein F, Link TM
(2005). Gender differences in trabecular bone architecture of the distal radius
assessed with magnetic resonance imaging and implications for mechanical
competence. Osteoporos Int. 2005 Sep; 16(9):1124–33.
78. Hunter DJ, Niu J, Zhang Y, Nevitt MC, Xu L, Lui LY, Yu W, Aliabadi P,
Buchanan TS, Felson DT (2005). Knee height, knee pain, and knee osteoarthritis:
the Beijing Osteoarthritis Study. Arthritis Rheum. 2005 May; 52 (5):1418-23.
79. Ipach I, Mittag F, Walter C, Syha R, Wolf P,Kluba T (2013). The prevalence of
acetabular anomalies associated with pistol-grip-deformity in osteoarthritic hips.
Orthop Traumatol Surg Res. 2013 Feb; 99 (1):37-45.
80. Javaid MK, Lane NE, Mackey DC, Lui LY, Arden NK, Beck TJ, Hochberg MC,
Nevitt MC (2009). Changes in proximal femoral mineral geometry precede the
onset of radiographic hip osteoarthritis: The study of osteoporotic fractures.
Arthritis Rheum. 2009 Jul; 60 (7):2028-36.
81. Jiang L, Rong J, Wang Y, Hu F, Bao C, Li X, Zhao Y (2011). The relationship
between body mass index and hip osteoarthritis: a systematic review and meta-
analysis. Joint Bone Spine. 2011 Mar; 78 (2):150-5.
64
82. Johnell O, Kanis JA, Oden A, Johansson H, De Laet C, Delmas P, et al (2005).
Predictive value of BMD for hip and other fractures. J Bone Miner Res 2005 Jul;
20(7):1185-94.
83. Johnson JK, Renner JB, Dahners LE (2012). Anteroposterior thickening of the
femoral neck with aging decreases the "offset" in men. Am J Sports Med. 2012
Oct; 40 (10):2213-7.
84. Jones G, Nguyen T, Sambrook PN, Lord SR, Kelly PJ, Eisman JA (1995)
Osteoarthritis, bone density, postural stability, and osteoporotic fractures: a
population based study. J Rheumatol. 1995 May; 22 (5):921-5.
85. Jones et al (2015). Proximal femoral fracture, Radiopaedia. Accessed from
<http://radiopaedia.org/articles/proximal-femoral-fracture> (Accessed 8th July,
2015).
86. Kanis JA, Glüer CC (2000). An update on the diagnosis and assessment of
osteoporosis with densitometry. Committee of Scientific Advisors, International
Osteoporosis Foundation. Osteoporos Int. 2000; 11 (3):192–202.
87. Kanis JA, McCloskey EV, Johansson H, Cooper C, Rizzoli R, Reginster JY
(2013). European guidance for the diagnosis and management of osteoporosis in
postmenopausal women. Osteoporos Int. 2013 Jan; 24(1):23-57.
88. Kawcak CE, McIlwraith CW, Norrdin RW, Park RD, James SP (2001). The role
of subchondral bone in joint disease: a review. Equine Vet J. 2001 Mar; 33(2):120-
6.
89. Kellgren JH, Lawrence JS (1957). Radiological assessment of osteo-arthritis.
Ann.Rheum.Dis. 1957 Dec; 16 (4):494-502.
65
90. Kinds MB, Bartels LW, Marijnissen ACA, Vincken KL, Viergever MA, Lafeber
FPJG, de Jong HWAM (2011). Feasibility of bone density evaluation using plain
digital radiography. Osteoarthritis Cartilage. 2011 Nov; 19 (11):1343-8
91. Kneejointsurgery.com (2015), Accessed from
<http://www.kneejointsurgery.com/articular-cartilage/ >, Accessed (24 July,
2015)
92. Knothe Tate ML, Adamson JR, Tami AE, Bauer TW (2004). The Osteocyte. Int
J Biochem Cell Biol. 2004 Jan; 36(1):1-8.
93. Kobayashi S, Takahashi HE, Ito A, Saito N, Nawata M, Horiuchi H, Ohta H, Ito
A, Iorio R, Yamamoto N, Takaokaa K (2003). Trabecular minimodeling in human
iliac bone. Bone. 2003 Feb; 32(2): 163–9.
94. Kolta S, Paratte S, Amphoux T, Persohn S, Campana S, Skalli W, Paternotte S,
Argenson JN, Bouler JM, Gagey O, Roux C (2012). Bone texture analysis of
human femurs using a new device (BMA™) improves failure load prediction.
Osteoporos Int. 2012 Apr; 23 (4):1311-6.
95. Korpelainen R, Korpelainen J, Heikkinen J, Väänänen K, Keinänen-
Kiukaanniemi S (2003). Lifestyle factors are associated with osteoporosis in lean
women but not in normal and overweight women: a population-based cohort study
of 1222 women. Osteoporos Int 2003 Jan; 14(1):34–43.
96. Kraus VB, Feng S, Wang SC, White S, Ainslie M, Brett A, Holmes A, Charles
HC (2009). Trabecular Morphometry by Fractal Signature Analysis Is a Novel
Marker of Osteoarthritis Progression. Arthritis Rheum. 2009 Dec; 60 (12): 3711–
22.
97. Kraus VB, Feng S, Wang S, White S, Ainslie M, Graverand MP, Brett A, Eckstein
F, Hunter DJ, Lane NE, Taljanovic MS, Schnitzer T, Charles HC (2013).
Subchondral bone trabecular integrity predicts and changes concurrently with
66
radiographic and magnetic resonance imaging-determined knee osteoarthritis
progression. Arthritis Rheum. 2013 Jul; 65 (7):1812-21.
98. Lammentausta E, Hakulinen MA, Jurvelin JS, Nieminen MT (2006). Prediction
of mechanical properties of trabecular bone using quantitative MRI. Phys Med
Biol. 2006 Dec 7; 51(23): 6187-98.
99. Lane NE (2007). Clinical practice. Osteoarthritis of the hip. N Engl J Med. 2007
Oct 4; 357 (14):1413-21.
100. Le Corroller T, Halgrin J, Pithioux M, Guenoun D, Chabrand P, Champsaur P
(2012). Combination of texture analysis and bone mineral density improves the
prediction of fracture load in human femurs. Osteoporos Int. 2012 Jan; 23 (1):163-
9.
101. Li G, Yin J, Gao J, Cheng TS, Pavlos NJ, Zhang C, Zheng MH (2013).
Subchondral bone in osteoarthritis: insight into risk factors and microstructural
changes. Arthritis Res Ther. 2013; 15(6): 223.
102. Lindsay R, Cosman F, Zhou H, Bostrom M, Shen V, Cruz J, Nieves JW, Dempster
DW (2006). A novel tetracycline labeling schedule for longitudinal evaluation of
the short-term effects of anabolic therapy with a single iliac crest biopsy: early
actions of teriparatide. J Bone Miner Res 2006 Mar; 21 (3): 366 –73.
103. Liu B, Balkwill A, Banks E, Cooper C, Green J, Beral V (2007). Relationship of
height, weight and body mass index to the risk of hip and knee replacements in
middle-aged women. Rheumatology (Oxford). 2007 May; 46 (5): 861–7.
104. Lohmander LS, Gerhardsson de Verdier M, Rollof J, Nilsson PM, Engström G
(2009). Incidence of severe knee and hip osteoarthritis in relation to different
measures of body mass: a population-based prospective cohort study. Ann Rheum
Dis. 2009 Apr; 68(4):490-6.
67
105. Looker AC, Wahner HW, Dunn WL, Calvo MS, Harris TB,Heyse SP, Johnston
CC Jr, Lindsay R (1998). Updated data on proximal femur bone mineral levels of
US adults. Osteoporos Int. 1998; 8 (5):468–489.
106. Madry H, van Dijk CN, Mueller-Gerbl M (2010). The basic science of the
subchondral bone. Knee Surg Sports Traumatol Arthrosc 2010, 18:419–433.
107. Marshall D, Johnell O, Wedel H (1996). Meta-analysis of how well measures of
bone mineral density predict occurrence of osteoporotic fractures. BMJ. 1996 May
18; 312 (7041):1254-9.
108. Mautalen CA, Vega EM, Einhorn TA (1996). Are the etiologies of cervical and
trochanteric hip fractures different? Bone. 1996 Mar; 18(3 Suppl):133S-137S.
109. Merle C, Waldstein W, Gregory JS, Goodyear SR, Aspden RM, Aldinger PR,
Murray DW, Gill HS (2014). How many different types of femora are there in
primary hip osteoarthritis? An active shape modeling study. J Orthop Res. 2014
Mar; 32 (3):413-22.
110. Messent EA, Ward RJ, Tonkin CJ, Buckland-Wright C (2005a). Cancellous bone
differences between knees with early, definite and advanced joint space loss; a
comparative quantitative macroradiographic study. Osteoarthritis Cartilage. 2005
Jan; 13(1):39-47.
111. Messent EA, Buckland-Wright JC, Blake GM (2005b). Fractal analysis of
trabecular bone in knee osteoarthritis (OA) is a more sensitive marker of disease
status than bone mineral density (BMD). Calcif Tissue Int. 2005 Jun; 76 (6):419-
25.
112. Messent EA, Ward RJ, Tonkin CJ, Buckland-Wright C (2005c). Tibial cancellous
bone changes in patients with knee osteoarthritis. A short-term longitudinal study
68
using Fractal Signature Analysis. Osteoarthritis Cartilage. 2005 Jun; 13 (6):463-
70.
113. Messent EA, Ward RJ, Tonkin CJ, Buckland-Wright C (2006). Differences in
trabecular structure between knees with and without osteoarthritis quantified by
macro and standard radiography, respectively. Osteoarthritis Cartilage. 2006 Dec
14 (12): 1302-5.
114. Messent EA, Ward RJ, Tonkin CJ, Buckland-Wright C (2007). Osteophytes,
juxta-articular radiolucencies and cancellous bone changes in the proximal tibia
of patients with knee osteoarthritis. Osteoarthritis Cartilage. 2007 Feb; 15 (2):179-
86.
115. Milz S, Putz R (1994). Quantitative morphology of the subchondral plate of the
tibial plateau. J Anat. 1994 Aug; 185 (Pt 1):103-10.
116. Moore RJ, Fazzalari NL, Manthey BA, Vernon-Roberts B (1994). The
relationship between head-neck-shaft angle, calcar width, articular cartilage
thickness and bone volume in arthrosis of the hip. Br J Rheumatol. 1994 May; 33
(5):432-6.
117. Müller D, Link TM, Monetti R, Bauer J, Boehm H, Seifert-Klauss V, et al (2006).
The 3D-based scaling index algorithm: a new structure measure to analyze
trabecular bone architecture in high-resolution MR images in vivo. Osteoporos
Int. 2006 Oct; 17(10):1483–93.
118. Murphy SB, Ganz R, Müller ME (1995). The prognosis in untreated dysplasia of
the hip. A study of radiographic factors that predict the outcome. J Bone Joint
Surg Am. 1995 Jul; 77 (7):985-9.
119. National Institute for Health and Care Exchange (NICE) (2014). The management
of hip fracture in adults; Issued: June 2011, last modified: March 2014; NICE
clinical guideline 124.
69
120. Nave R, Georgia State University (2015), Accessed from
<http://hyperphysics.phy-astr.gsu.edu/hbase/ems3.html#c4> (Accessed 12
October 2015).
121. Nevitt MC, Lane NE, Scott JC, Hochberg MC, Pressman AR, Genant HK,
Cummings SR (1995). Radiographic osteoarthritis of the hip and bone mineral
density. The Study of Osteoporotic Fractures Research Group. Arthritis Rheum.
1995 Jul; 38 (7):907-16.
122. Newton CD, University of Pennsylvania (2015), Etiology, classification, and
diagnosis of fractures, Accessed from
<http://cal.vet.upenn.edu/projects/saortho/chapter_11/11mast.htm> (Accessed
8th July, 2015).
123. Ojala T, Pietikainen M, Harwood D (1996). A comparative study of texture
measures with classification based on feature distributions. Pattern Recogn 1996;
29:51-9.
124. Oliveria SA, Felson DT, Cirillo PA, Reed JI, Walker AM (1999). Body weight,
body mass index, and incident symptomatic osteoarthritis of the hand, hip, and
knee. Epidemiology. 1999 Mar; 10(2):161-6.
125. Papaloucas CD, Ward RJ, Tonkin CJ, Buckland-Wright C (2005). Cancellous
bone changes in hip osteoarthritis: a short-term longitudinal study using fractal
signature analysis. Osteoarthritis Cartilage. 2005 Nov; 13(11): 998-1003.
126. Parfitt AM (2002). Targeted and nontargeted bone remodeling: relationship to
basic multicellular unit origination and progression. Bone. 2002 Jan; 30(1):5-7.
127. Partanen J, Jämsä T, Pekka Jalovaara (2001). Influence of the upper femur and
pelvic geometry on the risk and type of hip fractures. J Bone Miner Res. 2001
Aug; 16(8): 1540-6.
70
128. Patton MS, Duthie RA, Sutherland AG (2006). Proximal femoral geometry and
hip fractures. Acta Orthop Belg. 2006 Jan; 72 (1):51-4.
129. Peacock M, Turner CH, Liu G, Manatunga AK, Timmerman L, Johnston CC Jr
(1995). Better discrimination of hip fracture using bone density, geometry and
architecture. Osteoporos Int. 1995 May; 5 (3):167-73.
130. Plotkin LI, Aguirre JI, Kousteni S, Manolagas SC, Bellido T (2005).
Bisphosphonates and estrogens inhibit osteocyte apoptosis via distinct molecular
mechanisms downstream of extracellular signal-regulated kinase activation. J Biol
Chem. 2005 Feb 25; 280 (8).
131. Podsiadlo P, Dahl L, Englund M, Lohmander LS, Stachowiak GW (2008).
Differences in trabecular bone texture between knees with and without
radiographic osteoarthritis detected by fractal methods. Osteoarthritis Cartilage.
2008 Mar; 16 (3):323-9.
132. Preidler KW, White LS, Tashkin J, McDaniel CO, Brossmann J, Andresen R,
Sartoris D (1997). Dual-energy X-ray absorptiometric densitometry in
osteoarthritis of the hip. Influence of secondary bone remodeling of the femoral
neck. Acta Radiol. 1997 Jul; 38 (4 Pt 1):539-42.
133. Pulkkinen P, Partanen J, Jalovaara P, Jämsä T (2004). Combination of bone
mineral density and upper femur geometry improves the prediction of hip fracture.
Osteoporos Int. 2004 Apr; 15 (4): 274–280.
134. Pulkkinen P, Partanen J, Jalovaara P, Jämsä T (2010). BMD T-score discriminates
trochanteric fractures from unfractured controls, whereas geometry discriminates
cervical fracture cases from unfractured controls of similar BMD. Osteoporos Int.
2010 Jul; 21 (7):1269-76.
135. Pulkkinen P, Partanen J, Jalovaara P, Nieminen MT, Jämsä T (2011).
Combination of radiograph-based trabecular and geometrical parameters can
71
discriminate cervical hip fractures from controls in individuals with BMD in non-
osteoporotic range. Bone. 2011 Aug; 49 (2):290-4.
136. Rachidi M, Marchadier A, Gadois C, Lespessailles E, Chappard C, Benhamou CL
(2008). Laws' masks descriptors applied to bone texture analysis: an innovative
and discriminant tool in osteoporosis. Skeletal Radiol. 2008 Jun; 37 (6):541-8.
137. Raisz LG (1999). Physiology and pathophysiology of bone remodelling. Clin
Chem. 1999 Aug; 45(8 Pt 2):1353-8.
138. Reijman M, Hazes JM, Pols HA, Koes BW, Bierma-Zeinstra SM (2005).
Acetabular dysplasia predicts incident osteoarthritis of the hip: the Rotterdam
study. Arthritis Rheum. 2005 Mar; 52 (3):787-93.
139. Reikerås O, Høiseth A (1982). Femoral neck angles in osteoarthritis of the hip.
Acta Orthop Scand.1982 Oct; 53 (5):781-4.
140. Schott AM, Cormier C, Hans D, Favier F, Hausherr E, Dargent-Molina P, Delmas
PD, Ribot C, Sebert JL, Breart G, Meunier PJ (1998). How hip and whole-body
bone mineral density predict hip fracture in elderly women: the EPIDOS
Prospective Study. Osteoporos Int. 1998; 8 (3):247-54.
141. Schott AM, Hans D, Duboeuf F, Dargent-Molina P, Hajri T, Bréart G, Meunier
PJ; EPIDOS Study Group (2005). Quantitative ultrasound parameters as well as
bone mineral density are better predictors of trochanteric than cervical hip
fractures in elderly women. Results from the EPIDOS study. Bone. 2005 Dec; 37
(6):858-63.
142. Schouten JS, van den Ouweland FA, Valkenburg HA (1992). A 12 year follow up
study in the general population on prognostic factors of cartilage loss in
osteoarthritis of the knee. Ann Rheum Dis. 1992 Aug; 51(8):932-7.
72
143. Schuit SC, van der Klift M, Weel AE, de Laet CE, Burger H, Seeman E, et al
(2004). Fracture incidence and association with bone mineral density in elderly
men and women: the Rotterdam Study. Bone. 2004 Jan;34 (1):195-202
144. Schwartz AV, Vittinghoff E, Bauer D C et al. (2011) Association of BMD and
FRAX score with risk of fracture in older adults with type 2 diabetes. JAMA. 2011
Jun 1; 305 (21) 2184–92.
145. Seeley RR, Stephen TD, Tate P (1998). Human Anatomy and Physiology 4ed,
Mcgraw-Hill College, Columbus, OH, USA, pp 154-179.
146. Showalter C, Clymer BD, Richmond B, Powell K (2006). Three-dimensional
texture analysis of cancellous bone cores evaluated at clinical CT resolutions.
Osteoporos Int. 2006 Feb; 17 (2):259–66.
147. Spector TD, Hart DJ, Doyle DV (1994). Incidence and progression of
osteoarthritis in women with unilateral knee disease in the general population: the
effect of obesity. Ann Rheum Dis. 1994 Sep; 53(9):565-8.
148. Stone KL, Seeley DG, Lui LY, Cauley JA, Ensrud K, Browner WS, Nevitt MC,
Cummings SR; Osteoporotic Fractures Research Group (2003). BMD at multiple
sites and risk of fracture of multiple types: long-term results from the Study of
Osteoporotic Fractures. J Bone Miner Res. 2003 Nov; 18 (11):1947-54.
149. Suri S, Walsh DA (2012). Osteochondral alterations in osteoarthritis. Bone. 2012
Aug; 51(2):204-11.
150. Thevenot J, Hirvasniemi J, Pulkkinen P, Määttä M, Korpelainen R, Saarakkala S,
Jämsä T (2014). Assessment of Risk of Femoral Neck Fracture with Radiographic
Texture Parameters: A Retrospective Study. Radiology. 2014 Jul; 272 (1):184-91.
73
151. Turmezei TD, Treece GM, Gee AH, Poole KE (2013). Cortical thickness mapping
of the proximal femur: towards a new imaging biomarker of hip osteoarthritis.
Osteoarthritis and Cartilage, Volume 21, S194.
152. Tutorvista (2015), Accessed from
<http://www.tutorvista.com/content/physics/physics-ii/electricity/heating-
effect.php> (Accesssed 21 Octorber 2015).
153. Ubara Y, Fushimi T, Tagami T, Sawa N, Hoshino J, Yokota M, Kaitori H,
Takemoto F, Hara S (2003). Histomorphometric features of bone in patients with
primary and secondary hyperparathyroidism. Kidney Int 63:1809 –1816, 2003.
154. Ubara Y, Tagami T, Nakanishi S, Sawa N, Hoshino J, Suwabe T, Kaitori H,
Takemoto F, Hara S, Takaichi K (2005). Significance of minimodeling in dialysis
patients with adynamic bone disease. Kidney Int. 2005 Aug; 68 (2):833-39.
155. University of Glasgow (2015), Accessed from
<http://www.gla.ac.uk/t4/~fbls/files/fab/tutorial/generic/bone3.html> (Accessed
24 July 2015).
156. University of Sheffield (2015), Accessed from <https://www.shef.ac.uk/FRAX/>
(Accessed 24 July 2015).
157. Vestergaard P, Rejnmark L, Mosekilde L (2009). Osteoarthritis and risk of
fractures. Calcif Tissue Int. 2009 Apr; 84(4):249-56.
158. Vokes TJ, Giger ML, Chinander MR, Karrison TG, Favus MJ, Dixon LB (2006).
Radiographic texture analysis of densitometer-generated calcaneus images
differentiates postmenopausal women with and without fractures. Osteoporos Int.
2006 Oct; 17(10):1472–82.
74
159. Walker WT, Kawcak CE, Hill AE (2013). Medial femoral condyle morphometrics
and subchondral bone density patterns in thoroughbred racehorses. Am J Vet Res.
2013 May; 74 (5):691-9.
160. WHO (1994). WHO Technical Report Series 843 (1994). Assessment of fracture
risk and its application to screening for postmenopausal steoporosis.
161. WHO (2004). WHO scientific group on the assessment of osteoporosis at primary
health care level.
162. Wilkie JR, Giger ML, Chinander MR, Engh CA, Hopper RH Jr, Martell JM
(2008). Temporal radiographic texture analysis in the detection of periprosthetic
osteolysis. Med Phys. 2008 Jan; 35 (1):377-87.
163. Witt CM, Jena S, Brinkhaus B, Liecker B, Wegscheider K, Willich SN (2006).
Acupuncture in patients with osteoarthritis of the knee or hip: a randomized,
controlled trial with an additional nonrandomized arm. Arthritis Rheum. 2006
Nov; 54(11):3485-93.
164. Woloszynski T, Podsiadlo P, Stachowiak GW, Kurzynski M, Lohmander LS,
Englund M (2012). Prediction of progression of radiographic knee osteoarthritis
using tibial trabecular bone texture. Arthritis Rheum. 2012 Mar; 64(3):688-95.
165. Wolski M, Podsiadlo P, Stachowiak GW, Lohmander LS, Englund M (2010).
Differences in trabecular bone texture between knees with and without
radiographic osteoarthritis detected by directional fractal signature method.
Osteoarthritis Cartilage. 2010 May; 18 (5):684-90.