A real-time multi-threaded system to detect patient motion in SPECT imaging using multiple optical cameras

Patient motion, which causes artifacts in reconstructed images, can be a serious problem in SPECT imaging. If patient motion can be detected and quantified, the reconstruction algorithm can compensate for the motion. In previous work, we described a prototype system for tracking patient motion. In this paper, we present a real-time multithreaded visual tracking system (VTS) that will be suitable for deployment in clinical trials. The VTS tracks patients using multiple video images and image processing techniques, calculating patient motion in three-dimensional space. High performance is achieved by acquiring and processing images in parallel. Our system will incorporate up to five optical cameras monitoring the patient. Each camera is associated with a thread of control within the VTS. Thus, video images can be acquired asynchronously. As images are acquired, they are stored in buffers until needed. Another processing thread is responsible for requesting images from a specific time, and processing those images as a stereo set. The SPECT system operates in list mode. In this mode, all detected events are stored in a long list together with an event timestamp. Thus, it is possible to temporally register the time-stamped images with the detected activity in order to rebin the SPECT data to compensate for patient motion