Timing Neural Network Implementation in Monolithic PET detector

In Time-Of-Flight Positron Emission Tomography, enhancing coincidence time resolution (CTR) and sensitivity is essential for increasing the signal-to-noise ratio (SNR). While monolithic detectors have no spatial resolution restrictions, achieving excellent CTR continues to be challenging. Previous studies have employed machine learning techniques to improve CTR). In this work, we investigate the application of neural networks to predict the CTR using only experimental data and analog SiPMs between a monolithic PET detector and a pixelated reference detector. Traditional CTR estimation methods are subject to noise and uncertainty-causing factors, making neural networks a more accurate and reliable approach. Our results show that machine learning algorithms can improve timing resolution in monolithic crystals by decreasing CTR by 200%. Machine learning techniques in data processing have shown significant improvements in position estimation and are beginning to show promise in the timing domain. These findings highlight the potential advantages of incorporating machine learning techniques into PET data processing for improved image quality and SNR gain.