Image QualityImage Quality
Image quality can be described by Image quality can be described by five characteristicsfive characteristics– Contrast resolutionContrast resolution– Spatial resolutionSpatial resolution– Image noiseImage noise– Linearity/DistortionLinearity/Distortion– Uniformity/ArtifactsUniformity/Artifacts
What influences image What influences image quality?quality?
X-ray beam characteristicsX-ray beam characteristics DoseDose Transmissivity of the subject/Attenuation Transmissivity of the subject/Attenuation
characteristicscharacteristics Slice thicknessSlice thickness ScatterScatter Efficiency of analog-to-digital conversionEfficiency of analog-to-digital conversion Pixel sizePixel size Reconstruction algorithmReconstruction algorithm Display resolutionDisplay resolution
Image QualityImage Quality
Specification of image quality is Specification of image quality is usually very subjective and is usually very subjective and is described by such generic terms as described by such generic terms as detail, sharpness, or blurring.detail, sharpness, or blurring.– E.g. “the images of the patient appear E.g. “the images of the patient appear
blurry because she wouldn’t hold her blurry because she wouldn’t hold her breath”breath”
Image QualityImage Quality
Image quality cannot be represented Image quality cannot be represented by a single numberby a single number– There isn’t one all encompassing There isn’t one all encompassing
method that can accurately measure method that can accurately measure image qualityimage quality
Image quality can be represented by Image quality can be represented by several numbersseveral numbers
Image quality can be described Image quality can be described numerically by noise, linearity, and numerically by noise, linearity, and uniformityuniformity
Contrast ResolutionContrast Resolution
Contrast resolution describes the Contrast resolution describes the property of distinguishing between property of distinguishing between similar tissues, for example gray-similar tissues, for example gray-white matter in the brain, or white matter in the brain, or differentiating between the liver and differentiating between the liver and spleen.spleen.
Plain film radiography tends to lack Plain film radiography tends to lack contrast resolution (it has great contrast resolution (it has great spatial resolution however)spatial resolution however)
Plain Film vs. CTPlain Film vs. CTContrast ResolutionContrast Resolution
Examine the following imagesExamine the following images The liver is very hard to distinguish by The liver is very hard to distinguish by
looking at the radiographylooking at the radiography The CT does a much better job at The CT does a much better job at
demonstrating anatomical structures demonstrating anatomical structures associated with the liverassociated with the liver
This is because liver tissue is similar in This is because liver tissue is similar in density and CT is superior at detecting density and CT is superior at detecting objects with similar contrast valuesobjects with similar contrast values
Contrast ResolutionContrast Resolution
Determined by tissue atomic number Determined by tissue atomic number (Z), mass density, and electron density(Z), mass density, and electron density
Tissues with large difference in atomic Tissues with large difference in atomic number will have high contrastnumber will have high contrast– Lead (atomic number 82)Lead (atomic number 82)– Oxygen (atomic number 8)Oxygen (atomic number 8)
These items have high contrast (basically These items have high contrast (basically appearing white and black on a radiograph)appearing white and black on a radiograph)
Contrast ResolutionContrast Resolution
Tissues with large differences in mass Tissues with large differences in mass density will also have high contrastdensity will also have high contrast– Density is a measure of mass per unit of Density is a measure of mass per unit of
volumevolume– The SI unit is kg/mThe SI unit is kg/m3 3 but most people use but most people use
g/cmg/cm33 because it is more convenient because it is more convenient Sodium = 968 kg per cubic meterSodium = 968 kg per cubic meter Lead = 11340 kg per cubic meterLead = 11340 kg per cubic meter
– There would be a large difference in contrast There would be a large difference in contrast between these two elementsbetween these two elements
Contrast ResolutionContrast Resolution
During radiographic imaging, contrast During radiographic imaging, contrast resolution is improved with reduced resolution is improved with reduced scatter radiation and lower kVpscatter radiation and lower kVp– Scatter only contributes to noise and Scatter only contributes to noise and
decreases overall image qualitydecreases overall image quality– When kVp is increased the resultant When kVp is increased the resultant
image has a long scale of contrast image has a long scale of contrast (appears very grey with little (appears very grey with little differentiation between tissue types)differentiation between tissue types)
Contrast ResolutionContrast Resolution
CT employs a high kVp because this CT employs a high kVp because this helps to minimize patient dose by helps to minimize patient dose by reducing the amount of radiation that reducing the amount of radiation that is absorbed the patientis absorbed the patient
CT is able to maintain contrast CT is able to maintain contrast resolution because of the narrow x-resolution because of the narrow x-ray beam collimation which provides ray beam collimation which provides excellent scatter radiation rejectionexcellent scatter radiation rejection
Contrast ResolutionContrast Resolution
The larger the dynamic range, for The larger the dynamic range, for example 1024 versus 512, the less example 1024 versus 512, the less contrast in the imagecontrast in the image
At low spatial frequencies the modulation At low spatial frequencies the modulation transfer function (MTF) is a measure of transfer function (MTF) is a measure of contrast resolutioncontrast resolution
Contrast resolution is the ability to image Contrast resolution is the ability to image adjacent tissues having a similar mass adjacent tissues having a similar mass density and effective atomic numberdensity and effective atomic number
Contrast ResolutionContrast Resolution
Contrast resolution is improved by Contrast resolution is improved by using higher mAusing higher mA
Contrast resolution is improved by Contrast resolution is improved by imaging thicker slices imaging thicker slices
Contrast resolution is improved when Contrast resolution is improved when imaging smaller patientsimaging smaller patients
Contrast resolution is better with low Contrast resolution is better with low noise imagersnoise imagers
Contrast ResolutionContrast Resolution
Contrast resolution is improved with Contrast resolution is improved with larger field of view (FOV) and smaller larger field of view (FOV) and smaller matrix size, hence, larger pixelsmatrix size, hence, larger pixels
Contrast resolution can be improved Contrast resolution can be improved with a smoothing reconstruction filterwith a smoothing reconstruction filter
Spatial ResolutionSpatial Resolution
Spatial resolution refers to the ability Spatial resolution refers to the ability to faithfully reproduce small objects to faithfully reproduce small objects having high subject contrasthaving high subject contrast
The bone – soft tissue interface The bone – soft tissue interface represents very high subject contrastrepresents very high subject contrast
The liver – spleen interface represents The liver – spleen interface represents very low subject contrastvery low subject contrast
Spatial resolution is often described by Spatial resolution is often described by the degree of blurring in an imagethe degree of blurring in an image
Spatial ResolutionSpatial Resolution
A bone – soft tissue interface will be A bone – soft tissue interface will be very sharp and is described as a high very sharp and is described as a high spatial frequency interfacespatial frequency interface
High spatial frequency objects are High spatial frequency objects are more difficult to image than low more difficult to image than low spatial frequency objectsspatial frequency objects
Small, high contrast objects are more Small, high contrast objects are more difficult to image than large, low difficult to image than large, low contrast objectscontrast objects
Spatial ResolutionSpatial Resolution
Larger pixel size results in poorer Larger pixel size results in poorer spatial resolutionspatial resolution
Lower subject contrast results in Lower subject contrast results in poorer spatial resolutionpoorer spatial resolution
Larger detector size results in poorer Larger detector size results in poorer spatial resolutionspatial resolution
Larger prepatient and postpatient Larger prepatient and postpatient collimation results in more scatter collimation results in more scatter radiationradiation
Scatter RadiationScatter Radiation
Scatter radiation reduces contrast and Scatter radiation reduces contrast and results in less contrast resolutionresults in less contrast resolution
A smaller x-ray tube focal spot A smaller x-ray tube focal spot improves spatial resolution because of improves spatial resolution because of the sharper image projection, not the the sharper image projection, not the geometry of a shadowgeometry of a shadow
Point response function (PRF) is one Point response function (PRF) is one method of evaluating spatial method of evaluating spatial resolutionresolution
Spatial ResolutionSpatial Resolution
Edge response function (ERF) is one Edge response function (ERF) is one method of evaluating spatial resolutionmethod of evaluating spatial resolution
Line spread function (LSF) is one Line spread function (LSF) is one method of evaluating spatial resolutionmethod of evaluating spatial resolution
The Fourier transform (FT) is a The Fourier transform (FT) is a mathematical manipulation to convert mathematical manipulation to convert an intensity versus distance relationship an intensity versus distance relationship into an intensity versus 1/distance into an intensity versus 1/distance (spatial frequency) relationship(spatial frequency) relationship
Spatial ResolutionSpatial Resolution The FT of a PRF results in the MTFThe FT of a PRF results in the MTF The FT of an ERF results in the MTFThe FT of an ERF results in the MTF The FT of a LSF results in the MTFThe FT of a LSF results in the MTF Spatial resolution is best described by the Spatial resolution is best described by the
limiting spatial frequency (lp/cm)limiting spatial frequency (lp/cm) The MTF is obtained from the FT of the The MTF is obtained from the FT of the
PSF, LSR, or the ERFPSF, LSR, or the ERF The MTF is useful when evaluating The MTF is useful when evaluating
components of a system or comparing components of a system or comparing similar systemssimilar systems
Spatial ResolutionSpatial Resolution
Spatial resolution is improved with a Spatial resolution is improved with a smaller x-ray focal spot sizesmaller x-ray focal spot size
Spatial resolution is improved with Spatial resolution is improved with thinner section imagingthinner section imaging
Spatial resolution is improved with Spatial resolution is improved with narrow predetector collimationnarrow predetector collimation
Spatial resolution can be improved with Spatial resolution can be improved with the use of an edge enhancement the use of an edge enhancement (convolution) reconstruction algorithm(convolution) reconstruction algorithm
Spatial ResolutionSpatial Resolution
Spatial resolution is improved by Spatial resolution is improved by increasing the number of projection increasing the number of projection profiles acquired per scanprofiles acquired per scan
Spatial resolution is improved when small Spatial resolution is improved when small FOV or larger matrix size is employedFOV or larger matrix size is employed
At high spatial frequencies the MTF is a At high spatial frequencies the MTF is a measure of spatial resolutionmeasure of spatial resolution
The MTF is the principal means of The MTF is the principal means of expressing CT spatial resolutionexpressing CT spatial resolution
Spatial ResolutionSpatial Resolution
To understand MTF, one must first To understand MTF, one must first understand spatial frequency, which understand spatial frequency, which has units of line pairs/mm (lp/mm)has units of line pairs/mm (lp/mm)
One line and a line-sized interspace is One line and a line-sized interspace is a one line pair (lp)a one line pair (lp)
High spatial frequencies represent High spatial frequencies represent small objectssmall objects
Low spatial frequencies represent Low spatial frequencies represent large objectslarge objects
Spatial ResolutionSpatial Resolution
An MTF value of 1.0 represents an An MTF value of 1.0 represents an absolutely perfect imageabsolutely perfect image
As MTF value is reduced, image blur As MTF value is reduced, image blur increases and therefore image quality is increases and therefore image quality is reducedreduced
Usually, the spatial frequency at the .1 Usually, the spatial frequency at the .1 (10%) MTF is identified as the limiting (10%) MTF is identified as the limiting resolutionresolution
The total MTF of an imager is the product The total MTF of an imager is the product of component MTFsof component MTFs
Spatial ResolutionSpatial Resolution
Computed Tomography imagers are Computed Tomography imagers are capable of approximately 10 lp/cm (1 capable of approximately 10 lp/cm (1 lp/mm) in normal mode and up to lp/mm) in normal mode and up to approximately 20 lp/cm (2 lp/mm) in approximately 20 lp/cm (2 lp/mm) in the high resolution modethe high resolution mode
Z-axis resolution is better with spiral Z-axis resolution is better with spiral CT compared to conventional CTCT compared to conventional CT
Image NoiseImage Noise
Contrast resolution is limited by image Contrast resolution is limited by image noisenoise
Scatter radiation results in image noiseScatter radiation results in image noise An increase in slice thickness results in An increase in slice thickness results in
less noiseless noise An increase in slice thickness usually An increase in slice thickness usually
results in lower patient doseresults in lower patient dose An increase in pixel size results in less An increase in pixel size results in less
noisenoise
Image NoiseImage Noise
An increase in patient dose results in less An increase in patient dose results in less noisenoise
In statistics, noise is called standard In statistics, noise is called standard deviation deviation
High noise images appear blotchy, grainy, or High noise images appear blotchy, grainy, or spottyspotty
Low noise images appear very smoothLow noise images appear very smooth Noise in a CT image comes from the scanner Noise in a CT image comes from the scanner
electronics and the random nature of x-ray electronics and the random nature of x-ray interaction with a detectorinteraction with a detector
Image NoiseImage Noise
Anything that reduces CT noise will Anything that reduces CT noise will improve contrast resolutionimprove contrast resolution
Increased image noise at low mAs Increased image noise at low mAs can usually be accommodated by can usually be accommodated by using a low-pass convolution filterusing a low-pass convolution filter
Image noise can be reduced by using Image noise can be reduced by using 360 degree interpolation360 degree interpolation
LinearityLinearity
When water = 0, bone = 1000, and air = When water = 0, bone = 1000, and air = -1000, a CT imager exhibits perfect linearity-1000, a CT imager exhibits perfect linearity
The CT number for a given tissue is The CT number for a given tissue is determined by the x-ray linear attenuation determined by the x-ray linear attenuation coefficient (LAC)coefficient (LAC)
A plot of CT number versus LAC should be a A plot of CT number versus LAC should be a straight line passing through water straight line passing through water = 0= 0
LinearityLinearity
Good linearity is essential for quantitative Good linearity is essential for quantitative computed tomography (QCT)computed tomography (QCT)
Linearity is the ability of the CT image to Linearity is the ability of the CT image to assign the correct Hounsfield unit (HU) to assign the correct Hounsfield unit (HU) to a given tissuea given tissue
Linearity is monitored by imaging the 5-Linearity is monitored by imaging the 5-pin insert of the American Association of pin insert of the American Association of Physicists in Medicine (AAPM) and plotting Physicists in Medicine (AAPM) and plotting HU versus linear attenuation coefficientHU versus linear attenuation coefficient
UniformityUniformity
When a test object made of one When a test object made of one substance is imaged, the value of each substance is imaged, the value of each pixel should be the samepixel should be the same
Cupping is the reduction in CT number Cupping is the reduction in CT number toward the middle of a uniform test objecttoward the middle of a uniform test object
Peaking is the increase in CT number Peaking is the increase in CT number toward the middle of a uniform test objecttoward the middle of a uniform test object
Cupping and peaking are signs of poor Cupping and peaking are signs of poor image uniformityimage uniformity
UniformityUniformity
Image uniformity is essential for QCTImage uniformity is essential for QCT Uniformity is the ability of the CT Uniformity is the ability of the CT
imager to assign the same HU to a imager to assign the same HU to a uniform phantom (water) over the uniform phantom (water) over the entire FOVentire FOV
Contrast DetailContrast Detail
Contrast detail plots are instructive when Contrast detail plots are instructive when evaluating contrast resolution and spatial evaluating contrast resolution and spatial resolutionresolution
Contrast detail plots are useful when Contrast detail plots are useful when comparing different scanning protocolscomparing different scanning protocols
The low contrast region of the contrast The low contrast region of the contrast detail curve is noise limiteddetail curve is noise limited
The high contrast region of the contrast The high contrast region of the contrast detail curve is determined by the MTR of the detail curve is determined by the MTR of the imager componentsimager components