FPGA-Based Pulse Processing for Positron Emission Tomography

Data acquisition and pulse processing for positron emission tomography, or PET, has traditionally been an analog domain due to the unique performance requirements of the scanner hardware. Foremost on the performance requirements is the ability to timestamp the photodetector pulses with very high precision. This thesis proposes a new timing algorithm that leverages the increasing ADC sampling rates and FPGA computational power to eliminate the need for most of the analog circuits. This will simplify the design and possibly reduce the power of the data acquisition electronics. This thesis also enhances the pulse processing capabilities by adding an all-digital method of correcting for pulse pile-up. Additionally, we utilize the unique features of the FPGA to tune itself to specific pulse parameter of each photodetector in order to improve the pulse processing precision.;The algorithms developed in this thesis take advantage of the fairly consistent pulse shape of the photodetector pulses. The general idea is to fit a higher-resolution, predefined pulse to the lower-resolution data pulse from the photodetector. This "reference" pulse, which is built to have the same shape as the data pulses, is used to interpolate the start of the data pulse. This reference pulse is also utilized to separate pulses that are involved in pulse pile-up. The results demonstrate that a well-defined FPGA timing circuit can match the performance of an analog circuit while using relatively few FPGA resources. We also demonstrate that the computational power of FPGAs can be uses to make the signal processing more robust in the presence of pulse pile-up.