Automatic detection of scintillation light splashes using conventional and deep learning methods

Abstract Six methods for the automatic detection of scintillation light splashes in a portable gamma camera are compared. Each imaging frame might contain any number (including none), and location size each splash must be identified. For real-time imaging, identified characterised quickly with minimal processing overhead. The techniques compared on their ability to accurately determine number, position, splashes, reconstruct deposited energy within simulated data set known ground-truths. speed technique ease implementation also discussed. accuracy blob (light splash) localisation, Laplacian Gaussian approach was found provide most accurate estimation. However, its performance greatly relies appropriate tuning preprocessing parameters prior image analysis blobs frame. Deep learning approaches (Faster-RCNNs) performed significantly better than traditional algorithms terms predicting splash, did not require were more stable over range occupancies. Moreover, paper fine-tuned VGG16 based Faster-RCNN model detection, called DeepSplashSpotter (DSS).