Combination of a high resolution detector with small FOV and a low resolution detector with large FOV for high resolution and quantitative SPECT

• Published in: 2008 IEEE Nuclear Science Symposium Conference Record pp. 5229 - 5231
• Main Authors: Tsutomu Zeniya, Hiroshi Watabe, Hiroyuki Kudo, Yoshiyuki Hirano, Kotaro Minato, Hidehiro Iida
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01-10-2008
• Subjects: Collimators; Computer simulation; Detectors; Heart; Image reconstruction; Image resolution; Maximum likelihood detection; Performance evaluation; Radioactive materials; Thorax
• ISBN: 1424427142; 9781424427147
• ISSN: 1082-3654; 2577-0829
• DOI: 10.1109/NSSMIC.2008.4774412

SPECT using compact high resolution detector or pinhole collimator allows to image physiological functions with high resolution. However, when region-of-interest (ROI) is smaller than the object, the projection data are truncated due to radioisotope outside ROI. The truncation causes artifact and overestimation, which decrease quantitative accuracy. In theory, to eliminate the artifact and the overestimation due to truncation, the untruncated data from another large field-of-view (FOV) detector can be used even if the detector has low resolution. This study was aimed at evaluating feasibility of combination of a small FOV high resolution detector and a large FOV low resolution detector in clinical circumstance. This evaluation was performed by computer simulation with a numerical torso phantom. We tested whether the image in a selected small ROI (in this case, ROI was heart) can be obtained with high resolution and without artifact and overestimation. The small FOV detector with high resolution was with 1.14-mm resolution, 80-mm FOV and parallel collimator. The whole of heart was included in this FOV, but the surrounding area was truncated. The large FOV detector with low resolution has 9-mm resolution, 360-mm FOV and parallel collimator like conventional clinical SPECT. The untruncated projections including the whole of thorax were acquired by this detector. Gaussian noises were added to all projection data. Data from the small detector were reconstructed by maximum likelihood expectation maximization (MLEM) as iterative method, on the reconstruction matrix large enough to contain the whole of thorax. The reconstructed image from the large FOV detector was used as an initial image in iterative reconstruction. The image obtained by our proposed method had high resolution and the counts almost equivalent to that of original image in the small ROI. In conclusion, this result suggests feasibility of the combination of two detectors with small and large FOV to quantitatively obtain high-resolution image of a selected small ROI with clinical SPECT.