An Efficient Supervised Learning Technique for Tumour Detection and Analysis from MR Image Data Set

Abstract

Image mining plays a vital role in image analysis. It is a sub field of data mining technique and mainly focuses on knowledge discovery from image data sets. An Image mining technique uses mainly three major steps, image segmentation, detection and finally extraction of of information. In medical field image analysis for a medical image set such as MRI image data set, are always challenging for the researchers because medical image mining needs more accuracy in the mining results. Existing image mining methods encounters with several issues such as poor accuracy, higher detection time, and inaccurate tumour growth rate. In this research work we are presenting an efficient supervised learning method for tumour detection and analysis from MR image dataset. Proposed supervised learning method uses hybrid method. Initial it uses existing BWT method for data pre-processing and segmentation than later apply SVM+ PCA with object labelling method to extract final results for tumour image such as tumour size, type, growth rate. Existing BWT method with SVM and proposed BWT with SVM+ PCA are implemented over simulator MATLAB and various performance measuring parameters are calculated and experimental results analysis clearly shows that proposed method performs outstanding over existing BWT with SVM tumour detection method.