Performance Evaluation of Resolution/ Sensitivity Variable Slats-stack Collimator of C-SPECT Cardiac Scanner

Objectives: C-SPECT is a clinical human cardiac SPECT scanner developed for noninvasive diagnostic assessment of coronary artery disease. A unique feature of the system is its adaptive collimation, which allows for different tradeoffs between resolution, sensitivity, & field of view within a few seconds & without moving the patient. This is enabled by slit-slat collimator since the transverse & axial collimation are separable. The transverse collimation is a set of slits that can be moved along a conveyor system & the axial collimation is a set of adjustable slat-stacks. These slat stacks adjust by collapsing adjacent tungsten slats to double the gap between absorbing material. The intent of this study is to evaluate the performance of the slat-stacks (SS) in their two operating modes, compared both with theory & in terms of relative performance between modes.

Methods: Fourteen slat stacks were built using tungsten slats. The SS are mated one-to-one to the fourteen detector modules in the system following the same elliptical curvature as the slits. They were designed to provide isotropic resolution to a point source located at the center of the imaging volume. The SS modules consist of fixed (odd) & movable racks (even) holding tungsten plates parallel to each other & perpendicular to the detector modules. Air activated pneumatic linear actuators collapse the even plates to the odd-slats doubling the spacing between plates. In high-resolution (HR) mode, the pitch is 2.8 mm, whereas it is 5.6 mm in high-sensitivity (HS) mode. Point-source data were acquired over three axial positions & over a large portion of the axial field of view using computer-controlled stages within an axial plane. Those axial planes were approximately at 1/4, 1/2, & 3/4 axial SS height. Within a plane, the ^99mTc point source was translated through 7 positions in orthogonal stage axis with 30 mm increments for a total of 49 source locations. After each set of 49 acquisitions, the SS mode was toggled between HR & HS, with three data sets in each mode for each plane. The source was assayed at the beginning of each plane’s acquisitions & the scan duration at each location was extended to compensate for decay. The acquired data from grid point source points were fit to determine the parameters of the transformation between the scanner’s coordinate system & the stage coordinate so the position of the point source was then known in the scanner’s coordinate system. We calculated the sensitivity ratio at each source position & compared the experimental axial resolution with theoretical expectation.

Results: Sensitivity & resolution measurements are consistent with theoretical expectations, but vary by position. The sensitivity ranged from 5×10^-4 to 20×10^-4, with ratios of 1.8 (mean) & (+/- 0.1) (std. dev.). Similarly, the axial resolution ranged from 10.8 mm to 39.6 mm (full width at half maximum), with ratios of 1.61 (mean) & (+/-0.22) (std. dev.).

Conclusions: The sensitivity & resolution are consistent with expectations in high-resolution & high-sensitivity mode, with ratios near the theoretical limit of 2. The flexibility to change modes quickly is a great asset allowing the possibility of scout, dynamic, & high-resolution static imaging all in a single imaging position.