We present a complete system for motion correction in high resolution brain positron emission tomography (PET) imaging. The system is based on a compact structured light scanner mounted above the patient tunnel of the Siemens High Resolution Research Tomograph (HRRT) PET brain scanner. The structured light system is equipped with a near infrared diode and uses phase-shift interferometry (PSI) to compute 3D point clouds of the forehead of the patient. These 3D point clouds are progressively aligned to a reference surface, thereby giving the head pose changes. The estimated pose changes are used to reposition a sequence of reconstructed PET frames. To align the structured light system with the PET coordinate system, a novel registration algorithm based on the PET transmission scan and an initial surface has been developed. The performance of the complete setup has been evaluated using a custom-made phantom, based on a plastic mannequin head equipped with two positron-emitting line sources. Two experiments were performed. The first simulates rapid and short head movements, while the second simulates slow and continuous movements. In both cases, the system was able to produce PET scans with focused PET reconstructions. The system is nearly ready for clinical testing.
|Title of host publication||Image-Based Geometric Modeling and Mesh Generation|
|Editors||Yongjie (Jessica) Zhang|
|Publication status||Published - 2013|
|Series||Lecture Notes in Computational Vision and Biomechanics|