AI-Based Large-Area Nuclear Particle Track Analysis System

A Region-based Convolutional Neural Network (R-CNN) model was developed for automated analysis of alpha particle tracks. Columbia Resin 39 (CR-39) nuclear detectors were exposed to a 1-μCi Americium-241 ( 241 Am) alpha particle source and subsequently analyzed through large-area Scanning Electron Microscopy (SEM) imaging over the irradiated area. The R-CNN model was trained on annotated SEM images and achieved a total count accuracy of 95% with a precision of 94%, recall of 90%, and F1-score of 92% after preprocessing with Canny edge detection to enhance detection accuracy and computational efficiency. The automated analysis methods reduced analysis time by a factor of 1,000, requiring less than 10 minutes of user intervention. This approach represents a significant advancement in particle track analysis efficiency and accuracy for large-scale datasets.