Depth of interaction estimation using artificial neural network for continuous crystal PET detector

Artificial neural network has been proven to be an effective position estimator for a continuous crystal PET detector. It could be adapted to estimate DOT as well. Using our experimental setup, "8+8" signal readout scheme, and sideway irradiation, we acquired DOT network training data and test data. The selected DOT neural network was trained and evaluated. The test results show that the DOT networks could achieve DOT resolution about 2mm over the whole area of detector. By plane (x, y) network and DOT network, it is possible to achieve 3D position estimation from the single-end detected scintillation light distribution.