image quality in nuclear medicine

Image Quality in Nuclear Medicine

• Spatial Resolution, Contrast and Noise

• Evaluation of Detection and Observer Performance

• Quality Assurance of Imaging Instruments

Spatial Resolution

• Spatial resolution refers to the ability of imaging instrument to provide the sharpness or detail of images.

• Factors affecting spatial resolution include collimator resolution (the main factor in nuclear medicine). System sensitivity requires certain diameter of the collimator holes etc.

Spatial Resolution

• Factors affecting spatial resolution also include intrinsic resolution (due to the statistical variation which is photon energy dependent).

• Patient motion in respiratory or cardiac imaging causes image blurring

• Image display or recording system can also contribute the degrading of spatial resolution.

Methods for Evaluating Spatial Resolution

• Organ phantom measurement (qualitative) such as brain phantom.

• Bar phantom measurement (quantitative). There are a number of phantoms: four quadrant bar phantom, parallel-line phantom, orthogonal hole phantom.

• Point or Line Spread Functions.

Bar Phantom and Line Spread Function

• The smallest bar space distinguishable can be regarded as the system spatial resolution

• The FWHM of a point or line spread function is also a parameter for the spatial resolution.

• The FWHM of the spread function is about 1.4-2 times the resolvable bar pattern

Frequency Response Curve

• An graph of relative output amplitude versus frequency is an audio equipment’s frequency response curve.

• Inexpensive audio systems generally produce the “mid-range” audio frequencies (poor response in low and high frequencies)

Modulation Transfer Function

• An image system responds to the input image signal (spatial frequencies) is the modulation transfer function (contrast is the amplitude). MTF(ν)=Cout(ν)/Cin(ν) where Cin=(Imax-Imin)/(Imax+Imin)

• A typical nuclear medicine image system transfers lower image spatial frequencies.

Modulation Transfer Function

• Good low frequency response is needed to outline the coarse details of the image and is important for the presentation and detection of relatively large but low contrast lesion.

• Good high frequency response is necessary to portray fine details and sharp edges.

Modulation Transfer Function

• System’s spatial resolution can be described completely by the Modulation Transfer Function (MTF)

• MTF is obtained by mathematical analysis of the line or point spread functions

• Minimum visible bar patterns (frequency) can be related to the basic spatial frequency in MTF which drops below about 0.1.

Modulation Transfer Function

• MTF curves can be obtained for different components of an imaging system and the system MTF is then given by MTF(ν)=MTFi(ν)xMTFc(ν)

• The MTFs can be used for the comparison of different systems.

Contrast

• Image contrast refers to the differences in density (or intensity) in parts of the image.

• There are a number of factors that could affect the contrast such as the radiopharmaceuticals (high lesion-to-background uptake desirable)

• Film contrast (transparency has better contrast than Polaroid film) enhance both desire image contrast and noise.

Contrast

• Background count rates can reduce the image contrast substantially.

• Scattered radiation and septal penetration have the same effect of adding background to the image.

• Pulse height analyzer (narrow the window) could be used for the scatter rejection but there is a trade-off (decrease counts and increase noise)

Contrast

• Decreased contrast (background, scatter or septal penetration) results in poorer visibility of both large low contrast objects as well as fine details (all structures) in the image.

• Scatters add long tails to the spread function, suppress the low frequencies and shift the limiting high frequency in MTF.

Background Removal: Example

Noise

• Image noise may be either random or structured• Structured noise refers to non-random variation in

counting rate superimposed on and interfering with the perception of the structures of interest

• Structured noise may arise from the radionuclide distribution itself or caused by system artefacts.

• Random noise is caused by statistical variation of count rate and is very important factor in nuclear medicine.

Random Noise• Random noise is related to the information

density

• Information density is defined as the counts per unit area recorded

• Information density can be increased by increase count rate or imaging time

• Information density affects the minimum detectable size and contrast of lesions.

Noise vs Lesion Contrast• Noise contrast is the percentage standard

deviation of counts recorded in an area

• A 3-5 times the noise contrast is required for a lesion to be detectable

• Lesion contrast requirement increases as lesion size decrease

• Random noise may be the detection limiting factor for small, low contrast lesions.

Evaluation of Detection and Observer Performance

• Contrast-Details Studies (C-D)

• Receiver Operating Characteristics (ROC)

Contrast-Detail Studies

• A phantom with a set of objects of varying sizes and contrasts is used to acquire images from different systems. An observer is then given the images randomly to identify the smallest size visible at each level of contrast.

• Useful for comparing the detectability of both large low contrast lesions as well as small high-contrast lesions.

Receiver Operating Characteristic Studies

• A set of images is obtained with different image systems or techniques to be tested. A observer is asked to identify “positive” or “negative” with 4 levels of confidence.

• True positive fraction (TPF) and False positive fraction (FPF) are then calculated

• The ROC curves (TPF vs FPF) are then plotted for each system for comparison.

Quality Control Tests

• Operational checks and acceptance testing

• Quality parameters: uniformity, spatial linearity, spatial resolution, count rate response, system sensitivity, multi-window spatial registration, energy resolution etc.

• Quality control testing methods: common sources 99mTc and 57Co. Intrinsic and extrinsic

Acceptance Tests

• All parameters should be measured as the reference for the operational checks.

• Acceptance tests are more complicated than the routine tests but some of the tests are used for the operational checks.

Quality Control Program

• Physicists or Technologists are responsible. Quality control programs depend on every individual institution. There are general standard guidelines such NEMA standard.

• Current state-of-art computers and software programs are available for quality control purpose.