Imaging performance of two multiple-pinhole small-animal SPECT systems: Multiplexed vs. non-multiplexed data acquisition

Abstract

Increasing the number of pinholes in small-animal SPECT significantly improves its count sensitivity. When the detector(s) are small, however, overlapping of projections (multiplexing) from different pinholes is unavoidable and can amplify noise in reconstructed images. We have evaluated the performance of two multi-pinhole systems, one with and one without multiplexing, for a prototypical tumor-imaging task. We prepared seven beads (∼1.8-mm diameter) to mimic tumors labeled with Tc-99m. A uniform gelatin phantom was used to simulate normal background tissue. The tumor-to-normal tissue ratio was ∼6:1, and each bead contained ∼1 μCi. The first scanner, equipped with two 0.8-mm pinholes on each of three heads (HMS-0.8), acquired only non-overlapping projections. We also scanned the phantom using 1.6mm pinholes (HMS-1.6) The second scanner had 9 pinholes on each of four heads, and allowed multiplexing. To compensate for decay, the phantom was scanned for 50 min on HMS-0.8, 58 min on HMS-1.6, 82 min on NanoSPECT/CT with 1.4 mm pinholes (Nano-1.4), and 102 min with 1.0 mm pinholes (Nano-1.0). A total of 30 (24) angular projections were acquired with HMS (Nano); these were reconstructed using 10 OSEM subsets for HMS, and 4 subsets for Nano. The mean voxel value in each sphere, and the mean and standard deviation in a large VOI in the background, were used to compute the signal-to-noise ratio (SNR) and contrast of each bead. The relative noise in the background was also calculated. The systems with and without multiplexing yielded similar image quality and average bead SNR, especially for HMS-0.8 and Nano-1.0. Both systems yielded very similar SNR values, despite the fact that the multiplexed system acquired data using 36 pinholes, while the non-multiplexed system had only 6 pinholes. The multiplexed acquisition did not seem to adversely affect the image contrast of the spherical tumors.