Post date: Aug 18, 2009 2:39:10 AM
The good news is that results are pouring in from the registration algorithm that i deployed on Makoman1 and Glenoid (6677 registrations performed and counting). The bad news is that the accuracy looks terrible! As i looked deeper into the results, i quickly found the problem. For most of the vertebral bodies, i was using the wrong center of rotation! I had accidentally inverted the center positions for each spine so that the center of the cranial body was read in as the center of the caudal body. Obviously, this will have a significant effect on how rotations are applied to the body being registered, since rotations will be transformed into displacements. I have corrected this error and kicked off the registration again, so hopefully over the next few days, new and improved results will be cranked out.