An Efficient k-Means Clustering Algorithm: Analysis and Implementation
IEEE Transactions on Pattern Analysis and Machine Intelligence
Incorporating Non-rigid Registration into Expectation Maximization Algorithm to Segment MR Images
MICCAI '02 Proceedings of the 5th International Conference on Medical Image Computing and Computer-Assisted Intervention-Part I
Model-Based Brain and Tumor Segmentation
ICPR '02 Proceedings of the 16 th International Conference on Pattern Recognition (ICPR'02) Volume 1 - Volume 1
| Hi-index | 0.00 |
In this paper we describe a fully automated method for segmentation of cortical regions from 3D magnetic resonance images (MRI) of the brain using a pre-labelled anatomic atlas of the brain that guides the segmentation process. First, the method uses a linear transformation to establish a correspondence between the atlas and the subject to segment. Then, a classification algorithm performs a global segmentation to identify the anatomic tissues. Finally, an iterative process is performed between the tissue classified and nonlinear transformation to align locally the template with the tissue classified and to obtain spatial information to identify the interest structure; at each iteration the atlas is updated in order to improve the segmentation of the region of interest (e.g. the cortical region, sub cortical region and the ventricular system). This method can be used in the context of analysis and valuation of medical images. We validated the technique with real MR images of the brain and the results show that it can successfully segment structures of the brain, which has important medical applications.