Faculty of Engineering and Natural Sciences
A Generative Model for Image Segmentation based on Label Fusion
Dr. Mert Rory Sabuncu,
We propose a non-parametric, probabilistic model for the automatic segmentation of medical images, given a training set of images and corresponding label maps. The resulting inference algorithms we develop rely on pairwise registrations between the test image and individual training images. The training labels are then transferred to the test image and fused to compute a final segmentation of the test subject. Label fusion methods have been shown to yield accurate segmentation, since the use of multiple registrations captures greater inter-subject anatomical variability and improves robustness against occasional registration failures. To the best of our knowledge, this work presents the first comprehensive probabilistic framework that rigorously motivates label fusion as a segmentation approach. The proposed framework allows us to compare different label fusion algorithms theoretically and practically. In particular, recent label fusion or multi-atlas segmentation algorithms are interpreted as special cases of our framework.
In the set of experiments I will present in this talk, we use 39 brain MRI scans – with manually segmented white matter, cerebral cortex, ventricles and subcortical structures – to compare different label fusion algorithms and the widely-used Freesurfer whole-brain segmentation tool. Our results indicate that the proposed framework yields more accurate segmentation than Freesurfer and previous label fusion algorithms.
Bio: Since August 2006, Mert R. Sabuncu has been a post-doctoral researcher in Polina Golland's "medical vision" group at MIT's CSAIL. As of the fall of 2009, he is joining the faculty of
School and will be affiliated with the
Center of the
Hospital. Mert received his PhD degree from the Department of Electrical Engineering at
University in 2006. His research interests lie in biomedical image analysis, imaging genetics, signal/image processing, pattern recognition and artificial intelligence. His dissertation work was on information-theoretic multi-modal image registration. He used a minimum spanning tree based entropy estimation technique to design an efficient and fast rigid-body multi-modal image registration algorithm. His recent work focused on developing a data-driven population analysis approach, named iCluster, and exploring the theoretical and practical link between "machine learning" and "image registration."
September 25, 2009, 11:00, FENS G032