Image Characteristics and Effect of Acquisition Parameters in SPECT Imaging

Image Characteristics and Effect of Acquisition Parameters in SPECT Imaging

Objectives

  • Understand image characteristics of SPECT

  • Explain how acquisition parameters impact image quality

  • Describe how to improve SPECT image quality

  • Apply concepts of attenuation correction to SPECT images

Vocabulary

  • Background: Any non-specific radionuclide distribution

  • Noise: Statistical variations in counts registered

  • Attenuation: Detected radiation is reduced by scatter and absorption

  • Partial Volume Effect: Averaging of count density occurs within a voxel

  • Resolution Loss with Depth: Loss of resolution due to patient-collimator distance

  • Beam Hardening: Intensity of beam changes, with low energy photons being attenuated first

  • Scatter compensation: Removal or corrections for scatter of radioactivity within the patient

  • Resolution recovery: Correction for loss of resolution due to distance

  • Noise regularization: Removal of noise from images to improve contrast

SPECT Performance Characteristics

  • The issues affecting planar image quality also affect SPECT image quality due to the reconstruction process amplifying problems.

    • Background (nonspecific) activity

    • Noise (statistical variations)

    • Detector uniformity and linearity

    • Image contrast: How well cold lesions can be distinguished from hot areas

    • Loss of resolution with increasing detector distance: SPECT resolution is poorer than planar

    • Attenuation: Up to 25% of {Tc-99m} photons can be lost to attenuation

    • Compton scatter: Can account for up to 50% of detected gamma rays

    • Patient motion: Must review the raw 3D data and/or sinogram for accuracy

Factors Affecting SPECT Image Quality

  • Collimator

  • Energy peak/window

  • Matrix size: Options include 64 x 64 and 128 x 128

  • Image zoom

  • Detector orbit path: Circular, elliptical, or contoured options

  • Acquisition method: Step and shoot vs. continuous

  • Number of stops: General rule advises stops to be greater than or equal to the matrix size

  • Time per stop

  • Reconstruction method: Filtered back projection (FBP) vs. iterative techniques

Image Characteristics in SPECT

  • Background:

    • Small count density changes may appear significant in SPECT due to multiple angles of observation.

    • FBP spreads counts across the 3D image matrix, creating extra background noise.

    • Iterative methods avoid this added background.

  • Noise:

    • Each projection in SPECT has fewer counts than planar images, meaning noise constitutes a more significant percentage of total counts.

    • FBP reconstruction amplifies noise as high counts can spill over into neighboring pixels.

    • The selection of a low-pass filter is crucial as iterative techniques manage noise more effectively.

  • Resolution Loss with Depth:

    • SPECT experiences more significant resolution loss compared to planar imaging, partly due to the gamma camera's distance from the patient during acquisition.

    • Utilization of body contour orbits and long-bore collimators can enhance resolution.

  • Attenuation and Scatter:

    • As much as 25% of {Tc-99m} gamma rays may be lost due to attenuation; up to 50% can undergo Compton scattering.

    • Scatter disturbs image contrast by inaccurately placing counts.

  • Patient Motion:

    • Results in the same radioactivity source appearing in two different spatial locations, an effect visible on the sinogram image.

  • Partial Volume Effect:

    • Defined as the averaging that occurs within a voxel in a 3D imaging context due to finite volume sizes.

    • Limited resolution and smoothing issues in reconstruction lead to count spillover among adjacent slices.

    • Structures smaller than twice the imaging system's full-width at half-maximum (FWHM) can be influenced by surrounding areas.

Contrast in SPECT

  • Definition: The ability to detect differences in count density

    • Variability in a cold lesion (zero counts) versus a hot organ yields a contrast range from 0 (no difference) to 1 (maximum discrepancy).

  • SPECT Enhancements:

    • SPECT is noted for improving contrast by a factor of 4-7 compared to planar imaging.

    • This enhancement is a primary advantage of SPECT imaging.

Performance Measures in SPECT

  • Uniformity: Evaluated on a transverse slice of the 3D image matrix derived from a uniform cylindrical source activity concentration.

  • Spatial Resolution: Measured by tomographic acquisition of radioactivity from a line source, evaluating FWHM of the line spread function. SPECT spatial resolution is typically worse than planar imaging.

Effect of Acquisition Parameters on Image Quality

  • Choice of Collimator:

    • Higher-resolution collimators yield improved images; utilizing long-bore, slant-hole, or focused types enhances resolution for SPECT distances.

  • Energy Window:

    • There exists a compromise between maximizing total counts and minimizing the scatter's relative contribution.

  • Matrix Size:

    • Common options include 64x64, 128x128, and 256x256.

    • A finer matrix can lead to lower count densities and increase noise levels.

    • Using a 128x128 matrix is recommended when sufficient counts warrant it; consider the Nyquist frequency in relation to the anatomy under study.

  • Detector Orbit:

    • Step-and-shoot acquisition is preferable over continuous.

    • Elliptical or body contour paths are better than circular ones.

    • A 360º orbit typically produces better results than 180º except in cardiac studies.

    • Angular sampling (stops per rotation) should meet or exceed the matrix size for effective projections.

  • Zoom:

    • Applying a zoom factor effectively reduces pixel size and heightens Nyquist frequency, but intrinsic resolution limits zoom effectiveness beyond a factor of about 2.

  • Time per Stop:

    • Total study time should ideally remain under 30-45 minutes; considerations must be made around other acquisition parameters.

Reconstruction Methods

  • Filtered Back Projection (FBP):

    • Adjust low-pass filter parameters post-reconstruction to enhance image quality.

    • The Butterworth filter offers control for reducing graininess/noise via adjustments to critical frequency or order.

    • Key considerations include the Nyquist frequency and potential for artifacts such as ringing (oscillating patterns) and aliasing (incorrect interpretations due to oversampling frequencies).

  • Iterative Reconstruction:

    • Minimal operator intervention is required.

    • Quality of images is guided by the forward-projection step and is influenced by the model used for attenuation, scatter, and resolution loss assessment.

Improving SPECT Images

  • Attenuation and Scatter Considerations:

    • Gamma rays lost through photoelectric interactions decrease overall counts.

    • Attenuation Correction: An established 3D matrix compensates for attenuation to accurately portray radioactivity distribution.

    • Attenuation Map: This 3D representation highlights attenuation at each voxel.

  • Attenuation Correction Details:

    • Attenuation restricts detectors’ capabilities to distinguish between strong deep sources and weak shallow sources.

    • Underestimation of radioactivity may span from the edge to the body center (average factor of five).

    • A correction factor can be quantitatively applied to address tissue attenuation.

    • Simple methods average correction across different tissue types; advanced techniques utilize radiation sources (like x-ray or gamma ray) for real-time attenuation measurement.

  • Measuring Attenuation:

    • Narrow-beam geometry provides linear attenuation coefficient measurements without including scatter.

    • For {Tc-99m}, BCl = 0.15/cm

    • Broad-beam geometry is comparable to gamma camera settings but captures scattered photons, yielding BCl_{eff} = 0.12/cm for {Tc-99m}.

  • Beam Hardening Considerations:

    • Specify path length, types of tissue, and matter properties in broad-beam settings as more low-energy photons are absorbed compared to higher energy in matter interactions.

Patient-Specific Attenuation Maps

  • Creation of a 3D attenuation map can occur through simultaneous/sequential transmission and emission data acquisition, which utilizes an external radiation source and solves the attenuation equations for segmented tissues.

  • Attenuation maps enhance image quality and accuracy during iterative reconstruction processes.

Transmission-based Attenuation Correction with Rod Sources

  • Rod Sources: Using Gd-153 with notable photopeaks at 99 and 103 keV, consist of a geometry created from two opposing rod sources positioned 90º to each gamma camera.

  • During acquisition, rod sources shift across projections, and the corresponding energy window is adjusted accordingly, while a reference scan creates I(0) values needed for attenuation calculations.

CT-Generated Attenuation Maps

  • CT scans may replace rod-source systems by being affixed in front of gamma camera heads, creating attenuation maps.

    • It requires adjusting attenuation values from the CT scanner's energy to actual emission radionuclide energy.

    • Typically yields enhanced statistics and reduced noise compared to rods; however, it may expose patients to higher doses of radiation.

Incorporation into Iterative Reconstruction

  • The implementation of scatter compensation models effects in the iterative reconstruction’s forward-projection stage.

  • Resolution recovery incorporates collimator-lose effects as well.

Noise Regularization

  • The Bayesian inference model enforces constraints on the estimated 3D image matrix, acknowledging that:

    • Nuclear medicine images conform to Poisson statistics.

    • Areas with high density exhibit reduced noise.

    • Individual voxels cannot assume negative values.

    • Neighboring voxels ought to reflect reasonable similarity.

Technical Aspects of Implementation

  • Both resolution and noise regularization are foundational in producing effective attenuation and scatter compensations.

  • Gamma camera mechanisms should operate with higher precision tolerances, and routine quality control protocols must be rigidly adhered to.

Implementation Methods

  • Camera manufacturers have integrated noise regularization, resolution recovery, scatter correction, and attenuation correction, significantly enhancing image quality.

  • Software packages exist for introducing regularization improvements, and advanced imaging devices provide close focus on cardiac studies through small fields of view with pixelated detectors (use of CsI(Tl) or CZT crystals).

Clinical Benefits of Implementation

  • Enhancement of image quality by twofold may allow for reduced imaging times or dosages while maintaining quality levels.

  • Viability for stress-only myocardial perfusion imaging has been noted, along with anatomical correlation using CT-based attenuation maps.

Summary

  • The synthesis of attenuation correction, scatter compensation, noise regularization, and resolution recovery through iterative reconstruction yields substantial improvements in image quality.

  • Patients potentially benefit from reduced radiation doses or shorter imaging durations.