Dissimilarity‐based detection of schizophrenia

Abstract

In this article, a novel approach to schizophrenia classification using magnetic resonance images (MRI) is proposed. The presented method is based on dissimilarity-based classification techniques applied to morphological MRIs and diffusion-weighted images (DWI). Instead of working with features directly, pairwise dissimilar-ities between expert delineated regions of interest (ROIs) are considered as representations based on which learning and classification can be performed. Experiments are carried out on a set of 59 patients and 55 controls and several pairwise dissimilarity measurements are analyzed. We demonstrate that significant improvements can be obtained when combining over different ROIs and different dissimilar-ity measures. We show that combining ROIs using the dissimilarity-based representation, we achieve higher accuracies. The dissimilar-ity-based representation outperforms the feature-based representation in all cases. Best results are obtained by combining the two modalities. In summary, our contribution is threefold: (i) We introduce the usage of dissimilarity-based classification to schizophrenia detection and show that dissimilarity-based classification achieves better results than normal features, (ii) We use dissimilarity combination to achieve better accuracies when carefully selected ROIs and dissimi-larity measures are considered, and (iii) We show that by combining multiple modalities we can achieve even better results. V

Machine Learning Approaches: From Theory to Application in Schizophrenia

“…We applied the dissimilarity approach in two different studies [ 30 , 32 ], where 7 pairs of ROIs (for right and left hemispheres resp.) were considered, that is, amygdala, DLPFC, entorhinal cortex, Heschl's gyrus, hippocampus, superior temporal gyrus, and thalamus.…”

“…In Ulaş et al [ 30 ], gray value histograms and their probability density functions (pdfs) were obtained for each ROI in 124 subjects (64 with diagnosed schizophrenia plus 60 HC), from which a T1-weighted MRI sequence had been acquired, while in Ulaş et al [ 32 ] two different MRI modalities were used, that is, T1-weighted and DWI sequences, both acquired from a cohort of 114 subjects (59 with diagnosed schizophrenia plus 55 HC). Since ROIs were traced on T1-weighted scans, in the latter study a coregistration step was needed in order to properly realign each ROI from the T1-weighted space into the DWI space.…”

“…In both papers the number of dissimilarities matrices was given by the product of the number of ROIs (14), the number of dissimilarity measures (13), and the number of modalities considered (1 in [ 30 ] and 4 in [ 32 ]).…”

“…In Ulaş et al [ 32 ], in addition to the 1-nearest neighbour (NN) rule on the original dissimilarities and the linear SVM in dissimilarity space, linear SVM classifier on the original feature space was employed. Besides, in addition to classification performed at a single-ROI level and combining all considered ROIs, a further classification was obtained combining different MRI modalities.…”

“…To overcome this problem, dissimilarity vectors can be used instead of feature vectors. In this case, similarities measure is computed between each couple of subjects in the dataset and directly used as a feature in the SVM analysis; such a method is referred to as pairwise dissimilarity approach [ 30 ].…”