CT

Seeram Chapter 11:

Image Quality

SpatialResolution

ImageNoise

ContrastResolution

Artifacts

BeamCharacteristics

DoseSlice

Thickness

Scatter

DisplayResolution

ReconstructionAlgorithm

SubjectTransmissivity

Spatial ResolutionQuantifies image blurring“Ability to discriminate objects of varying

density a small distance apart against a uniform background”

Minimum separation required between two high contrast objects for them to be resolved as two objects

Spatial Resolution

Resolvable Object Size &Limiting Resolution

Smallest resolvable high contrast objectOften expressed as line pairs / cm“Pair” is one object + one space

OnePair

Resolvable Object Size:Limiting ResolutionSmallest resolvable high contrast object is

half the reciprocal of spatial frequencyExample:

Limited resolution = 15 line pairs per cmPair is 1/15th cmObject is half of pair

1/15th / 2 1/30th cm .033 cm 0.33 mm

1/15th cm

1/30th cm

Geometric Factors affectingSpatial Resolution

Focal spot sizedetector aperture widthslice thickness or collimationLess variation likely for thinner slices

attenuation variations within a voxel are averagedpartial volume effect

Geometric Factors affectingSpatial Resolution

focal spot - isocenter distance

focal spot - detector distance

Finite focalspot size

Geometric Unsharpness & CTDecreased spatial resolution if object blurred over several detectors

Detector aperture sizemust be < object for object to be resolved

Focal Spot

Detectors

Small Object to be

Imaged

Non-geometric Factorsaffecting Spatial Resolution

# of projectionsDisplay matrix size

512 X 512 pixels standardReconstruction algorithms

smoothing or enhancing of edges

Reconstruction Algorithm &Spatial ResolutionBack projecting blurs imageAlgorithms may be anatomically specific

Special algorithmsedge enhancementnoise reductionsmoothingsoft tissue or bone emphasis

Hi-Resolution CT TechniqueVery small slice thicknesses

1-2 mmHigh spatial frequency algorithms

increases resolution increases noise Noise can be offset by using higher doses

Optimized window / level settingsSmall field of view (FOV)

Known as “targeting”

Contrast Resolution

Ability of an imaging system to demonstrate small changes in tissue contrast

The difference in contrast necessary to resolve 2 large areas in image as separate structures

CT Contrast Resolution

Significantly better than radiographyCT can demonstrate very small differences in

density and atomic #

Radiography10%

CT<1%

This’ll be on your test. I guarantee it.

CT Contrast Resolution Depends Upon

reconstruction algorithmlow spatial frequency algorithm smooths

image Loss of spatial resolution Reduces noise

enhances perceptibility of low contrast lesions

image display

CT Contrast Resolution Depends on Noise

CT Contrast Resolution

Contrast depends on noise

Noise depends on # photons detected

# photons detected depends on …

# of Photons Detected Depends Upon

photon flux (x-ray technique)slice thicknesspatient sizeDetector efficiencyNote:

Good contrast resolution requires that detector sensitivity be capable of discriminating small differences in intensity

Small Contrast Difference Harder to Identify in Presence of Noise

CT Image NoiseFluctuation of CT

#’s in an image of uniform material (water)

Usually described as standard deviation of pixel values

CT Image Noise

Standard deviation of pixel values

(xi - xmean)2

Noise () = -------------------(n-1)

Xi = individual pixel valueXmean = average of all pixel values in ROIn =total # pixels in ROI

Noise Level

UnitsCT numbers (HU’s)

or% contrast

Noise Measurement in CT

Scan water phantomSelect regions of

interest (ROI)Take mean & standard

deviation in each regionStandard deviation

measures noise in ROI

CT Noise Levels Depend Upon

# detected photons quantum noise

matrix size (pixel size) slice thickness algorithm electronic noise scattered radiation object size Photon flux to

detectors…

Photon Flux to Detectors

Tube output flux (intensity) depends uponkVpmAsbeam filtration

Flux is combination of beam quality & quantity

Flux to detectors modified by patientLarger patient = less photons to detector

Slice ThicknessThinner slices mean

less scatter better contrast

less active detector area less photons detected More noise

To achieve equivalent noise with thinner slices, dose (technique factors) must be increased

Noise Levels in CT:Increasing slice width = less noise

BUTIncreasing slice width degrades spatial

resolutionless uniformity inside a larger pixelpartial volume effectpartial volume effect

2() = kT/(td3R)

Where is variance resulting from noisek is a conversion factor (constant)T is transmissivity (inverse of attenuation)t is slice thicknessd is pixel sizeR is dose

Noise Levels in CT:When dose increases, noise decreases

dose increases # detected photonsDoubling spatial resolution (2X lp/mm)

requires an 8X increase in dose for equivalent noiseSmaller voxels mean less radiation per

voxel

2() = kT/(td3R)

If slice thickness goes down by 2 Dose must go up by 2

To hold noise constant

Measurements of Image Quality

PSF = Point Spread FunctionLSF = Line Spread FunctionCTF = Contrast Transfer FunctionMTF = Modulation Traffic Function

Point Spread FunctionPSF

“Point” object imaged as circle due to blurring

Causesfinite focal spot sizefinite detector sizefinite matrix sizeFinite separation between object and

detector Ideally zero

Finite distance to focal spot Ideally infinite

Quantifying BlurringObject point becomes image circleDifficult to quantify total image circle size

difficult to identify beginning & end of object

Intensity

?

Quantifying BlurringFull Width at Half Maximum (FWHM)width of point spread

function at half its maximum value

Maximum value easy to identify

Half maximum value easy to identify

Easy to quantify width at half maximum

FWHM

Maximum

HalfMaximum

Line Spread FunctionLSF

Line object image blurredImage width larger than object width

Intensity

?

Contrast Response FunctionCTF or CRF

Measures contrast response of imaging system as function of spatial frequency

LowerFrequency

HigherFrequency

Loss of contrast between light and dark areas as bars & spaces get narrower. Bars & spaces blur into one another.

Contrast Response FunctionCTF or CRFBlurring causes loss of contrast

darks get lighterlights get darker

LowerFrequency

HigherFrequency

HigherContrast

LowerContrast

CT PhantomsAvailable from

CT manufacturerprivate phantom

manufacturersAmerican Association

of Physicists in Medicine AAPM

Measure• noise

spatial resolution• contrast resolution• slice thickness• dose

CT Spatial vs. Contrast ResolutionSpatial & contrast resolution interact

High contrast objects are easier to resolveOmprove one at the expense of the otherCan only improve both by increasing dose

Increasing object size

Increasing contrast

Contrast & DetailLarger objects easy to see even at low

contrast

Increasing object size

Increasing contrast

Contrast & DetailSmall objects only visible at high contrast

Increasing object size

Increasing contrast

Contrast – Detail RelationshipContrast vs. object diameter

less contrast means object must be larger to resolve

Difference in CT #

Object Diameter

Increasing object size

Increasing contrast

Visibility

Modulation Transfer FunctionMTF

Fraction of contrast reproduced as a function of frequency

RecordedContrast

(reduced by blur)frequency

MTF

1

0Contrast provided

to film

Freq. =line pairs / cm

50%

MTF

Can be derived frompoint spread functionline spread function

MTF = 1 meansall contrast reproduced at this frequency

MTF = 0 meansno contrast reproduced at this frequency

MTF

If MTF = 1all contrast reproduced at this frequency

RecordedContrast

Contrast providedto film

MTF

If MTF = 0.5half of contrast reproduced at this

frequency

RecordedContrast

Contrast providedto film

MTF

If MTF = 0no contrast reproduced at this frequency

RecordedContrast

Contrast providedto film

CT Number

Calculated from reconstructed pixel attenuation coefficient

t - W)CT # = 1000 X ------------

W

Where:ut = linear attenuation coefficient for tissue in pixeluW = linear attenuation coefficient for water

LinearityLinear relationship of CT #’s to

object linear attenuation coefficientsChecked with phantom of several

known materialsaverage CT # of each material

obtained from ROI analysisCompare CT #’s with known

coefficients

325

-100

50 -44

77

Cross-Field Uniformity Use uniform phantom (water) CT pixel values should be uniform

anyplace in image Take 5 ROI

1 center ROI4 corners ROI’s

Compare standard deviation between ROI’s

CT ArtifactsDistortionAreas where image

not faithful to subjectSources

patientimage processequipment

CT ArtifactsDistortion

Phantoms with evenly distributed objects

Preview!CT Artifacts: Causesmotionmetal & high-contrast

sharp edgesbeam hardeningpartial volume averagingsamplingdetectors

Motion Artifacts

Causes streaks in imageAlgorithms have trouble coping because of

inconsistent data

Artifacts: Abrupt High Contrast Changes Examples:

prostheses dental fillings surgical clips Electrodes bone

Metal absorbs all radiation in ray causes star-shaped artifact

Can be reduced by software

CT Artifacts:Beam Hardening

Increase in mean energy of polychromatic beam as it passes through patient

Can cause broad dark bands or streakscupping artifact

Reduced by beam hardening correction algorithms

CT Artifacts:Partial Volume EffectCT #’s based on linear attenuation coefficient for

tissue voxelsIf voxel non-uniform (contains several materials),

detection process will average

Partial Volume Effect

Can appear as incorrect densities streaks bands

Minimizing Use thinner slices

Image Artifacts:Ring Artifact in 3rd GenerationCauses

1 or more bad detectorssmall offset or gain

difference of 1 detector compared to neighbors detector calibration required

Reason: rays measured by a given detector are all tangent to same circle

Quality Control in CTPerformance tested at prescribed intervals

• spatial resolution

• contrast resolution

• noise

• slice width

Image Quality Tests

• kVp waveform

• average & standard deviation of water phantom CT #

• scatter & leakage