SPECT is an important technique for assessing myocardial perfusion. However SPECT suffers from low sensitivity because of the necessity for collimation. New systems have emerged with higher sensitivity for myocardial perfusion imaging, a primary market for SPECT. Most of these designs focus on imaging a region of interest around the heart. The design by Funk et al and Discovery series from GE use multiple pinholes. The advantage of pin-hole designs is that there are no moving parts, thus reducing manufacturing and servicing costs. We propose a novel method to improve resolution and/or sensitivity by using curved detectors fitted to pin-hole collimators. The use of a curved detector on each pin-hole results in a better resolution over that of a flat-detector attached to the same pin-hole, because of improved magnification. The improved resolution can then be swapped for sensitivity by increasing the pin-hole diameter, to obtain better sensitivity performance for similar resolution as the flat-detector. We mathematically derived expressions for average resolution of paraboloid, conical and spherical detectors, and calculated the sensitivity improvement keeping the same average resolution as the LEHR parallel collimators normally used for Cardiac SPECT. We also implemented an analytical ray-tracing based forward projector for the paraboloid detector with the model of the pin-hole resolution. The ray-tracing simulations corroborated our analytical results. We devised a stationary configuration of the pinholes where sets of pinholes are focused on the cardiac volume. Our preliminary design with the paraboloid-shaped detector behind each pin-hole, showed that the sensitivity improvement over the high-sensitive multi-pin-hole system with flat-detectors was 48-85% (depending upon the number pin-hole used). The sensitivity improvement of the curved detector over the clinical systems currently used for cardiac imaging was a factor of 7.4 to 9.3, with resolution similar to that of LEHR in the region of interest. Our hypothesis is that further improvements may be achieved by finding the optimum curved detector surface and most importantly, by modeling the pin-hole resolution and penetration and detector parallax effects in iterative reconstruction for compensation. Thus potentially there would be further resolution improvement to be traded to obtain improved sensitivity. Our overall goal is to thus formally optimize the detector surface and model the resolution and penetration and detector parallax effect within the reconstruction algorithm and evaluate the performance improvements for our proposed system.
The work proposed in this grant is distributed unevenly between the two years. As shown in the research plan, in the first year it is proposed that all the theoretical design and modeling (which are the PI's responsibility) will be finished. The implementation of the forward projector and reconstruction algorithm (which is the post-doctoral fellow's responsibility, under guidance and help from the PI) will be started. This encompasses Specific Aim 1 and significant parts of Specific Aim 2 and Specific Aim 3. In the second year the implementation of the algorithm will be finished and different designs will be evaluated and compared. The degree of effort of the PI and in particular the co- investigators is expected to be less for the second year. In the second year the PI will continue to guide and help the postdoctoral fellow in finishing the implementation of the reconstruction algorithm and evaluating the different designs and writing up the material for publication in peer-reviewed journals. The uneven-modular budget reflects this unevenness in work-plan. The PI's relative engagement would be 55%-44%. It is expected that advice from the co-Investigators'will be needed more in the first year of the grant. The post-doctoral fellow will be equally engages throughout the two years.
| Chan, Chung; Dey, Joyoni; Grobshtein, Yariv et al. (2016) The impact of system matrix dimension on small FOV SPECT reconstruction with truncated projections. Med Phys 43:213 |
