A novel automated approach for segmenting lateral ventricle in MR images of the brain using sparse representation classification and dictionary learning

Segmenting lateral ventricle in medical images plays an important role in medical diagnosis. The volume of lateral ventricle increases with age and it is an important indicator of Alzheimer´s, schizophrenia, and depressive disorders. In this article a new approach based on sparse representation and dictionary learning as a pre process of the existing active contour models for segmentation is introduced to automatically segment this area. In recent years utilizing methods for sparsely representing a signal over a given dictionary has gained considerable attention by scholars. Applications of signal sparse representation varies from compression to denoising, classification and many more. This article expands this growing area of research into a new level, by introducing a new approach for segmenting MRI images utilizing sparse representation solutions. The method takes advantage of K-SVD dictionary learning algorithm to create two distinct over complete dictionaries for each class and it uses sparse representation classification (SRC) algorithm to sparsely represent the image as well as discriminating the two different classes in the image.