Anomaly detection through registration

We introduce a system that automatically segments and classifies features i n brain MRI volumes. It segments 144 structures of a 256 x 256 x 124 voxel image in 18 minutes on an SGI computer with four 194 MHz R10K processors. T he algorithm uses an atlas, a hand-segmented and classified MRI of a normal brain, which is warped in 3-D using a hierarchical deformable matching alg orithm until it closely matches the subject. This customized atlas contains the segmentation and classification of the subject's anatomical structures . We have conducted tests on 139 MRIs of normal brains, and 3 MRIs and 1 CT of brains with pathologies. We present qualitative and quantitative evalua tions of the system's performance. Combined with domain knowledge, the regi stration algorithm is capable of detecting asymmetries and abnormal variati ons in the subject's data that indicate the existence and location of patho logies. (C) 1999 Pattern Recogntion Society. Published by Elsevier Science Ltd. All rights reserved.