An Automated System for Tumor Detection Using Details MRI Image

Abstract: Brain tumor in human occur when abnormal cells collect within the brain. This paper proposed an appearance based method for Tumor detection using MRI. In this appearance based method wavelet is followed by the PCA. The wavelet is embedded with the PCA (Principal Component Analysis) to overcome the limitations of PCA and improve the tumor recognition rate. DWT is applied on image to extract the important information (Detail image) and to remove the irrelevant information of the MRI. Then PCA is applied on the detail MRI image to find the uncorrelated feature vector. After extracting features from the MRI, classification is done by k-nearest neighbor (KNN). The Classification rates for the proposed feature extraction techniques are analyzed with different classifier pairs. To compare these feature extraction techniques for correct tumor recognition, an experiment is conducted on datasets which is collect from Anant hospital Jabalpur. The experimental result shows the acceptable rate for detection of tumor. Keywords: Wavelet; MRI; Detail MRI; Principal Component Analysis; k-nearest neighbor