A Multi-Time over Threshold Data Acquisition System for Silicon Photomultipliers based Positron Emission Tomography Imaging

Abstract

Recent developments in the area of Positron Emission Tomography (PET) detectors using Silicon Photomultipliers (SiPMs) have demonstrated the feasibility of higher resolution PET scanners due to a significant reduction in the detector form factor. However, reduced size implies a corresponding increase in the detector density, resulting in a proportional rise in the number of channels interfacing a SiPM array with the digital backend. In addition, the fast timing response of the SiPMs requires very high-resolution data acquisition systems with sampling speeds of multiple Gigasamples/sec, resulting in considerably high power and area budget. The main focus of this work is to explore analog and mixed signal circuit design techniques to enable power and area optimized data acquisition system design. As part of this work, three different readout systems were implemented in CMOS silicon. The first readout chip was fabricated in STMicroelectronics 130 nm CMOS process. The main focus of the ASIC was to reduce the number of channels in the backend using row column channel combining scheme and employ threshold detection technique to reduce dark noise contribution across the readout channels. The second generation of the readout chip was fabricated in TSMC 65nm Low Power process with a modified front-end interface design. The third generation of the ASIC, the main focus of this thesis, presents a chip level implementation of a Time-over-Threshold based data acquisition system for PET Imaging systems, which will help to replace the high-speed digitizers with a power and area optimized design. This was fabricated in TSMC 65nm Low Power process flavor. With multiple thresholds, this chip enables multiple time stamps on the fast rising edges, thereby resulting in an improved timing resolution for the system. These time stamps can be used to develop better algorithms for image reconstruction, using various statistical estimation techniques and machine learning algorithms. Also, multiple ToT helps in improving the tradeoff between timing and energy resolution. This thesis addresses different challenges associated with implementing a high resolution ToT system. This new implementation will allow more compact scanner/detector designs without having to sacrifice any of the data fidelity, reduce cost, and increase reliability.