Brain Tumor Segmentation using ANN technique and Wavelet Feature Extraction

  • T.Chithambaram, Dr.K.Perumal
Keywords: MRI, brain tumor, ANN, SVM, segmentation, competitive learning

Abstract

Brain tumor is a life threatening disease and requires computer based diagnostic support. The paper proposes a novel wavelet based feature set and introduced unsupervised Artificial Neural Network (UANN) to segment brain tumor from magnetic resonance images (MRI). In this paper, competitive layers are introduced in ANN to become unsupervised. Further, the proposed method introduces wavelet based feature set that automatically detects abnormal images and avoids the process of abnormal image detection. Existing works uses ANN to solve the abnormal image detection, tumor type and grade categorization. The proposed work adapts UANN to segment the timorous tissues from normal tissues. The final results are compared and analyzed against Support Vector Machine (SVM) classifier in terms of quantitative and qualitative measures. Dice similarity index measure (Dice) is used for quantitative analysis. The results proved the outstanding results of the proposed method UANN.

Published
2021-07-13
How to Cite
Dr.K.Perumal, T. (2021). Brain Tumor Segmentation using ANN technique and Wavelet Feature Extraction. Design Engineering, 2725- 2737. Retrieved from http://www.thedesignengineering.com/index.php/DE/article/view/2664
Section
Articles