Monolithic crystals on SiPM arrays in a prototype system with depth of interaction estimation

Increased sensitivity is one of the major challenges in novel detector development. While pixelated crystals are standard, the use of continuous crystals can provide increased active volume and thus sensitivity. Accurate estimation of the interaction positions of photons in continuous crystals requires a sophisticated algorithm. Additionally, including depth of interaction (DoI) in the reconstruction process can mitigate parallax effects near the edges of field of view, therefore DoI estimation is highly desirable. In this work, the interaction position coordinates in 3D are estimated adapting an already existing analytical model to our system with no additional hardware, training data sets or Monte Carlo simulations. The algorithm is implemented in a two-detector-head system comprised of continuous crystals coupled to SiPM arrays. Acquired data is stored in list-mode where the continuous nature of the interaction position is preserved. For image reconstruction, since the crystal is continuous, pre-computation of the system matrix elements would require discretization of the detected measurements. To avoid data discretization the system matrix elements are calculated on-the-fly. Spatial information is later on retained in the reconstruction algorithm by using a Siddon-like algorithm where the ray end-points correspond to the estimated interaction positions in continuous space, introducing the interaction position estimation uncertainty in the calculation of the system matrix elements. Reconstruction using continuous data is compared to reconstruction using artificially discretized data. Results show that the spatial resolution measured from a reconstructed point source using ML-EM obtained with a continuous crystal is superior to the spatial resolution obtained with a pixelated crystal.