Quantification of collagen architecture:3D thickness structure

Internship project

Biomedical Image Analysis & Tissue Engineering

October 2006 – January 2007

Desiree Abdurrachim

Supervisor: Bart terHaar Romeny & Mirjam Rubbens

Desiree Abdurrachim 2

• Introduction

• Blobness measure: a blob predictor

• Algorithm

• Results

• Conclusions

• Future works

Outline

Desiree Abdurrachim 3

The thickness of collagen fiber determines the strength of the engineered tissue [Billiar et.al]

Aim:

To quantify the thickness of the fiber

Introduction [1/2]

Desiree Abdurrachim 4

Introduction [2/2]• the cross-sectional area of the fiber is assumed to be a ci

rcle• Implemented in Mathematica• Tested to the real data: 4 week-old strained B2_63_loc1• Limited for the fiber parallel to x-axis

x-axis

Desiree Abdurrachim 5

• Introduction

• Blobness measure: a blob predictor

• Algorithm

• Results

• Conclusions

• Future works

Outline

Desiree Abdurrachim 6

Blobness measure [1/2]

A good detector for a blob structure:

• Higher measure for a better-fit scale

a 66.32 41.89

b 61.23 72.36

a

b

Desiree Abdurrachim 7

Blobness measure [2/2]• Experiment:

the relation between the real diameter and :

• Linear relation!diameter = -0.11 + 2.84

Desiree Abdurrachim 8

• Introduction

• Blobness measure: a blob predictor

• Algorithm

• Results

• Conclusions

• Future works

Outline

Desiree Abdurrachim 9

Algorithm [1/3]

Input: 3D CED image (256 x 256 x 81)

Extracted fiber

skeleton

Centre point

Fiber extraction

Skeleton extraction(thinning)

Blobness measure in cross-sectional area

Fiber indexing

Diameter prediction in each fiber

Desiree Abdurrachim 10

Algorithm [2/3]

Extracted fiber

Cross-sectional areas: blobness measure =0.8 to 7, step =0.2

Fiber extraction

Skeleton extraction(thinning)

Blobness measure in cross-sectional area

Fiber indexing

Diameter prediction in each fiber

Desiree Abdurrachim 11

Algorithm [3/3] – check!

Follow this fiber:

• Consists of 6 cross-sectional area (parallel to x-axis)

• Observe three slices:

• By observation : 6 – 8 pixels

• By blobness measure:

Desiree Abdurrachim 12

• Introduction

• Blobness measure: a blob predictor

• Algorithm

• Results

• Conclusions

• Future works

Outline

Desiree Abdurrachim 13

Results [1/2]• Distribution of in the calculated fibers

• What happens in = 0.8?• What happens in = 7?

Desiree Abdurrachim 14

Results [2/2]• Calibration: scale pixels (1 pixel = 0.78 m) • A set of diameters in one fiber• Distribution:

• Diameter : 1.6 m to 6 m– heart leaflet diameter : 20 m & depends on age [Baer et.al]– smaller value: due to the young tissue (4 weeks)

Desiree Abdurrachim 15

• Introduction

• Blobness measure: a blob predictor

• Algorithm

• Results

• Conclusions

• Future works

Outline

Desiree Abdurrachim 16

Conclusions• A framework to quantify the thickness of the fibers has

been successfully developed• Circular assumption blobness measure• Implemented in Mathematica • Tested to the real data• The experiment has shown promising results• Limitations:

– No phantom validation– Quantification only for fibers parallel to x-axis– Not able to measure < 2 pixels (=1.6 m)

Desiree Abdurrachim 17

Future work• Phantom validation• Inclusion of the fibers with other orientations

- gauge coordinate (v and w are eigenvectors)

- modification of the blobness measure:• Experiments on various data (strained-unstrained, you

ng-mature)

Desiree Abdurrachim 18

References

• [Billiar et. al]Billiar K.L. Sacks M.S., Biaxial mechanical properties of the native and glutaraldehyde-treated aortic valve cusp: Part II-A Structural constitutive model. Journal of Biomedical Engineering., 122:137-335, 2000

• [Baer et. al]Baer E., Cassidy James J., and Hiltner A. Hierarchical structure of collagen composite systems: lesson from biology. Pure & Appl. Chem., Vol. 63, No.7, pp. 961-973, 1991.

Desiree Abdurrachim 19

References• Phantom validation

• dd

Thank you!