This article presented a framework for classifying structural brain Magnetic Resonance (MR) images by integrating deformation-based morphometry with machine learning. A moral description of the exciting anatomy is first obtained with a high-dimensional mass pre-serving prototype warping process, resulting in maps of tissue density forming the volumetric tissues of the local tissue. Regions with high associations between the volume of tissue and diagnosis (medical) variables are derived using the water shield segmenting algorithm to ensure robustness for outliers, taking into account a geographic smoothness of the correlation map, which has been calculated by cross-validation technique. A volume increase algorithm is then used in those regions for extracting regional volumetric features, which are used to select more discriminatory features by using the feature selection technique based on support vector creator (SVM) parameters, depending on the effect on the superior boundary of the generalization error. Finally, the SVM classification is used for the best functionality and checked using a cross-validation exit technique. In MR brain pictures, not just high classification precision (91.8% female and 90.8% male) of stable monitors and psychotic patients indicates strong stability concerning the number of selected features and the scaled SVM core. © 2023 Elsevier B.V., All rights reserved.