Support Vector Machine-Based Brain Image Classification and Its Application in Diagnosis of Mental Diseases

Abstract

Since the training samples are usually limited in medical image segmentation, it is usually difficult for traditional model classification methods to obtain good results, the purpose of this paper was to deeply study support vector machine (SVM) method and its application in the diagnosis of mental diseases in medical image segmentation. This paper adopted SVM method which had a good classification performance in the small sample, nonlinear and high-dimensional feature space to study the characteristics of medical image segmentation. Results indicated that the S type function-based fuzzy support vector machine method had more accurate effect than the traditional support vector machine. A conclusion can be drawn that the method of determining the degree of membership based on tightness can effectively distinguish outliers or noisy samples from valid samples in the sample set relative to distance-based membership degree.