CLSTM-KF reconstruction method for a low-activity moving radiation source localization and tracking with a coded-aperture gamma camera

摘要

Background Accurate localization of a low-activity moving radiation source plays an important role in nuclear security and safety. The coded-aperture gamma camera is generally applied to detect a radiation source, but its reconstruction methods may have some limitations when the radiation source is motional and weak. Purpose The purpose of this paper is to improve the quality of the reconstruction images and the localization accuracy when detecting a low-activity moving radiation source with a gamma camera. Method The CLSTM-KF method consists of the CLSTM network and the Kalman filter. The CLSTM network is applied to improve the CNR of reconstruction images by making an adaptive superposition for sequential reconstruction images decoded by the correlation analysis method. After the CLSTM network, a series of sequential positions would be filtered by the Kalman filter. Results By comparing with the traditional methods of the gamma camera, the CLSTM-KF method performs well in improving both the CNR of reconstruction images and the localization accuracy. Moreover, the computation time of the CLSTM-KF method can also meet the application requirements. Conclusion In summary, the CLSTM-KF method provides a better choice than the traditional methods in locating and tracking a low-activity moving radiation source.