CVPR10 Clustering
An Automatic Unsupervised Classification of MR Images in Alzheimer's Disease
Abstract
Image-analysis methods play an important role in help- ing detect brain changes in and diagnosis of Alzheimer’s Disease (AD). In this paper, we propose an automatic unsu- pervised classification approach to distinguish brain mag- netic resonance (MR) images of AD patients from those of elderly normal controls. The symmetric log-domain diffeo- morphic demons algorithm, with the properties of symme- try and invertibility, is used to compute the pair-wise reg- istration, whose deformation field is then used to calculate the Riemannian distance between them. The spectral em- bedding algorithm is performed based on the Riemannian distance matrix to project images onto a low-dimensional space where each image is represented as a point and its neighboring points correspond to images of high anatom- ical similarity. Finally, the quick shift clustering method is employed in the embedded space to partition the dataset into subgroups. The experiments using the proposed method show very good performance for clustering images into AD and normal aging, using the Clinical Dementia Rat- ing (CDR) scale as a comparison.
Reference and Preprint
X. Long and C. Wyatt, 
An Automatic Unsupervised Classification of MR Images in Alzheimer's Disease, 23rd IEEE Conference on Computer Vision and Pattern Recognition. San Francisco, USA. June 13-18, 2010. [Spotlight Video Presentation - starts at 20:44]
Code and Example Data
The following archive contains the source code, scripts, and instructions for replicating the results in the paper: SpectralClusteringPackage.tar.gz.
Tested Configurations
Linux-x86 (GCC 3.4.3)
Comments and Bug Reports