Initial Investigations of Pinhole Numbers and Projection Views in Multi-pinhole Brain SPECT

SPECT is a powerful tool for diagnosing neurological diseases such as Alzheimer’s disease (AD) and Parkinson’s disease (PD), while multi-pinhole (MPH) SPECT could offer superior resolution-detection efficiency trade-off. Here we aim to investigate the relationship of pinhole numbers and projection views in MPH brain SPECT for a conventional dual-head SPECT scanner. The NCAT brain phantom was used to simulate 2 activity distributions, i.e.,
^99m
Tc-HMPAO for AD and
^99m
Tc-TRODAT for PD. The target system spatial resolution was set to 12 mm for a 200 mm FOV to cover the whole patient’s head. We used a 3D analytical MPH projector to generate 64, 32, 16, 8, 4 and 2 noise free projections for 1-15 pinholes and noisy projections for 1-20 pinholes respectively. The noise levels for different pinholes were modelled by scaling the clinical count level for 2 brain applications obtained from a LEHR parallel hole collimator. Noise free and noisy projections were reconstructed using the ML-EM algorithm up to data convergence. Normalized mean square error (NMSE) was calculated over the whole brain region for HMPAO and striatum region for TRODAT in both noise free and noisy simulation, respectively. Based on the noise-free NMSE analysis for 64 to 2 projections, the optimum pinhole number increased from 3 to 11 for HMPAO and from 1 to 10 for TRODAT. While for noisy data, the optimum pinhole number increased from 5 to 18 for HMPAO and from 2 to 13 for TRODAT for 64 to 2 projections. We conclude that more number of pinholes are needed as the number of projections decreases. More number of pinholes are needed for HMPAO as compared to TRODAT.