Effect of K-Fold Cross Validation on Mri Brain Images Using Support Vector Machine Algorithm
M. Jaya Lakshmi1, S. Nagaraja Rao2

1M. Jaya Lakshmi, Research Scholar, JNTUA, Department of ECE, G. Pullareddy Engineering College, Kurnool (Andhra Pradesh), India.
2Dr. S. Nagaraja Rao, Professor, Department of ECE, G. Pullareddy Engineering College, Kurnool (Andhra Pradesh), India.
Manuscript received on 24 April 2019 | Revised Manuscript received on 06 May 2019 | Manuscript Published on 17 May 2019 | PP: 301-307 | Volume-7 Issue-6S4 April 2019 | Retrieval Number: F10570476S419/2019©BEIESP
Open Access | Editorial and Publishing Policies | Cite | Mendeley | Indexing and Abstracting
© The Authors. Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open access article under the CC-BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/)

Abstract: Recently, exact detection of the cancerous tumor in brain images is a critical task, especially at the early stage of the diseases. Various investigators have used machine-learning methods for the computer-aided diagnosis (CAD) to detect the tumor. In this paper an accurate and an automatic CAD system frame work has been done for verifying, the effect of K-fold cross validation for different values of k. K-means the segmentation technique is in the initial phase of the framework and the image is pre-processed for feature extraction and feature reduction using 2D-DWT and PCA respectively. The reduced features are given to the machine learning algorithm called the kernel support vector machine to classify magnetic resonance images. The K-fold stratified cross validation scheme is utilized to simplify the ability of the suggested strategy. The proposed method uses the different fold cross validation schemes, it is found that the RBF type kernel achieves the maximum classification with k=5 for the given data set. This method of classification of MR brain images, can help radiologists to analyze whether the patient’s stage is normal or abnormal.
Keywords: Brain Tumor, Principal Component Analysis, Feature Extraction, Classification, Segmentation, Image De-noising, Principal Component Analysis (PCA), two-Dimensional Discrete Wavelet transform (2D- DWT), Kernel Support Vector Machine (KSVM).
Scope of the Article: Image Processing and Pattern Recognition