Component-based Point Spread Function Modeling for an Adaptive Multi-Pinhole SPECT with Full-Ring Detectors

We developed an adaptive whole-body SPECT/CT with a full detector ring and eight sets of multi-pinhole (MPH) collimators. Each MPH collimator was designed for specific clinical practices with optimized resolution-sensitivity tradeoff. Accurate modeling of the point spread function (PSF) is critical for delivering high resolution and quantitative images. In this work, we present a component-based method to accurately model the PSF and quantitatively assess its impact on the imaging performance. The PSF consists of the collimator and detector response function (CDRF) and the object dependent function (ODF). Firstly, the CDRF is modeled as a function of the isotope energy, incident angle and source location in FOV to account for the detector response and the penetration and scatter in pinhole material. The CDRF parameters is calibrated by a hybrid approach of Monte-Carlo simulation and experimental measurements. A point source of Tc-99m and I-131 (<;0.5mm) is precisely scanned in FOV in simulation and experiments and then the resulting projections are used for fitting and fine-tuning the CDRF model. Secondly, the ODF is modeled as a function of the μ-map, isotope energy, and source location to account for attenuation and scatter in the object. The object μ-map is derived from the CT images, and the latter two related ODF parameters is calibrated by scanning a point source in water phantoms with different diameters. Finally, the PSF for each voxel is calculated on-the-fly with the calibrated parameters in iterative reconstruction to generate images for evaluation. Comprehensive simulation and experimental studies are underway to assess the image quality in terms of spatial resolution, uniformity, contrast, recovery coefficients and quantitative accuracy.