Development of Pattern Recognition Software for Tracks of Ionizing Radiation In Medipix2-Based (TimePix) Pixel Detector Devices

The principal aim of our project is to develop an efficient pattern recognition tool for the automated identification and classification of tracks of ionizing radiation as measured by a TimePix version of the hybrid semiconductor Medipix2 pixel detector system. Such a software tool would have a number of applications including dosimeters to assess the risk of human exposure to radiation, and area monitors to characterize the general background radiation environment harmful to humans and electronic equipment. We are particularly interested in the development of the real-time analysis software needed to support an operational dosimeter that can assess the radiation environment during space missions. Our software development project makes use of data taken in beams of heavy ions at HIMAC (Heavy Ion Medical Accelerator Facility) in Chiba, Japan, including data from several different heavy ions with similar Linear Energy Transfers (LETs) for calibration purposes. We describe two modules of our pattern recognition tool: feature generation and classification. Our first module builds on a segmentation algorithm that identifies tracks from the pixel image assuming an approximately elliptical form that varies in size and degree of elongation based on multiple factors, including the particle species and angle of incidence. Determining the charge and energy of the particles creating each track is a particularly challenging task because different energy and charge incident particles can produce very similar patterns. Our classification module invokes different algorithms such as decision trees, support vector machines, and Bayesian classifiers.