Clinical usability of a compact high resolution detector for high resolution and quantitative SPECT imaging in a selected small ROI

SPECT using compact high resolution detector or pinhole collimator allows to image physiological functions with high spatial resolution. However, when field-of-view (FOV) is smaller than the object, the projection data are truncated by radioisotope outside FOV. The truncation causes artifact and overestimation, which decreases quantitative accuracy. Recently Defrise et al proposed a new truncation-compensated reconstruction method, that is, the truncated data can be successfully reconstructed by fulfilling following conditions. First, FOV contains zero or background counts outside the object as known value. Second, reconstructed image space is large enough to contain the whole support of the object. They demonstrated their theory by 2D X-ray CT simulation. This study was aimed at evaluating clinical-SPECT usability of a reconstructed image of a selected small region-of-interest (ROI) with the above Defrise’s method. This evaluation was performed by computer simulation with a numerical human brain phantom and a detector with 2-mm resolution, 48-mm FOV and a parallel collimator. The projection data were acquired including the area outside the brain. After adding Gaussian noise, the projection data were reconstructed by maximum likelihood expectation maximization (MLEM) method on the reconstruction matrix large enough to contain the whole support of the brain. This simulation showed that the truncation compensated reconstruction method could provide the image with high resolution and the counts almost equivalent to that of original image in the selected small ROI without the effect of truncation for human brain. In conclusion, this result suggests that a compact high resolution detector can be used for quantitatively reconstructing a selected small ROI with clinical SPECT camera. This technique can also use the pinhole collimator instead of the compact high resolution detector.