Tumor Detection in MRI Images using Modified Multi-level Otsu Thresholding (MLOT) and Cross-Correlation of Principle Components

Abstract

Tumors are normally classified by analyzing the magnetic resonance imaging (MRI) of the human brain. The proposed work has developed a programmed machine-based tumor recognition from MRI images. This work describes another strategy for the detection of tumors from patients’ brain MRI images by examining images using the multi-level otsu thresholding (MLOT) method. In this work, MRI of test patients and MRIs of the normal human brain compared using principal components and based on the differences measured, the tumor detection will be done. The normal human brain MRI's taken from CC-BY-SA standard SRI24 multichannel atlas of normal adult human brain structure. The proposed method incorporates with noise removal using discrete wave transform (DWT) filter and MRI Image preprocessing includes modified multi-level otsu thresholding (MLOT) and morphological operations like erosion and dilation and last step a cross-correlation of principal components based tumor classification performed. MATLAB 2018b has been used for designing and testing the proposed work, graphical user interface (GUI) has been developed for frontend users and the accuracy of detection is found better than other similar work.