A virtual dissection based registration to model patient-specific respiratory motion

This work is directed at reducing patient induced blurring in SPECT imaging due to breathing motion. As image resolution improves this breathing motion is becoming increasingly significant. Method: The NCAT phantom and an associated medical image processing package (RMDP) are used to obtain a breathing cycle of images (both CT and corresponding SPECT) and a full organ segmentation. A process termed ´virtual dissection´ is undertaken which sees individual organs extracted from the main images and independently registered (ICP). These individual registrations are reconciled, combined and used to obtain improved final images. Results: The results of the objective validation techniques are presented together with a comparison of processed and unprocessed images. Conclusion: Within the scope of the synthetic data used and for organs for which the assumption of near-rigid motion holds well the technique works. In the case of the ribs and lungs further development is needed.