Quantitative Processing of Cardiac Dynamic Patient Data With Slow Camera Rotation

Abstract

A study aimed at the extraction of kinetic information from clinical cardiac SPECT is presented. A dual head SPECT system with high resolution parallel-hole collimators was used for imaging the distribution and kinetics of 201Tl at rest and 99mTc-sestamibi at stress in the myocardium of patients. Dynamic acquisition was performed for each pharmaceutical by acquiring 48 tomographic projection sets over a 30 min period. Both the dynamic datasets and the summed dataset were reconstructed using a standard voxelized reconstruction approach and a new point cloud grid multi-resolution representation. ROI analysis was used to extract dynamic information from the scan data acquired 30 seconds or more after the start of acquisition. Since the dynamics of the initial injected bolus are too fast in order to obtain a consistent tomographic data set during the first 36 seconds of the scan, an estimate of the time activity curve for the blood input was obtained by applying factor analysis directly to the projections. Time basis functions and expansion coefficients for each image element in the point cloud representation were estimated from projections using conjugate gradient optimization.