SPARSE REPRESENTATION BASED CLASSIFICATION OF MR IMAGES OF BRAIN FOR ALZHEIMER’S DISEASE DIAGNOSIS

Vinith Rejathalal, Srinu Babu Mandalapu, V K Govindan

Abstract: High-dimensional pattern classification methods, like support vector machines (SVM), have been widely used for analysis of structural and functional brain images to assist the diagnosis of Alzheimer’s disease (AD). However, due to noise and small sample size of neuroimaging data, it is challenging to train a global classifier that can be robust enough to achieve good classification performance. In this paper, we evaluate the performance of the SVM and Dictionary learning method such as K-SVD by using a local patch based method. Dictionary Learning involves finding a basis set (dictionary) that best represents the data and finding a sparse representation in terms of the dictionary. We evaluated our method on 165 subjects (including 55 AD patients, 54 MCI and 56 normal controls) from Alzheimer’s disease Neuroimaging Initiative (ADNI) database using MR images. The K-SVD method gives better results than SVM by using local patch based method. By using K-SVD getting very promising performance when compared to SVM method for AD/MCI classification using MR images.

Keywords: Sparse representation; Feature extraction; dictionary learning; K-SVD; local patch

DOI: https://doi.org/10.15623/ijret.2014.0304131