An automated MRI brain image segmentation and tumor detection using SOM-clustering and Proximal Support Vector Machine classifier

Abstract

In recent days, image processing is an interesting research field and mainly the medical image processing is increasingly challenging field to process various medical image types. It is widely used in diagnosis of disease such as brain tumor, Cancer, Diabetes etc. and brain tumor is one such dangerous disease and currently moreover 600,000 people have this type of disease. Image segmentation is an important technique highly used to extract the suspicious parts from medical images such as MRI, CT scan, and Mammography etc. With this motivation in this work, SOM clustering is proposed for MRI brain image segmentation. Before the segmentation the Histogram Equalization is utilized for feature extraction which will improve the segmentation accuracy. After the segmentation process, the feature extraction using Gray Level Co-occurrence Matrix is utilized which avoids the formation of misclustered regions. The Principle Component Analysis (PCA) method is used for the feature selection to improve the classifier accuracy. An effective classifier Proximal Support Vector Machines (PSVM) is used to automatically detect the tumor from MRI brain image. This method is faster and computationally more efficient than the existing method SVM. While the SOM clustering with Histogram Equalization is a fast procedure for the segmentation of the whole volume and provides a way to model tissue classes, the PSVM-GLCM-PCA approach is a more robust scheme under noisy or bad intensity normalization conditions which produces better results using high resolution images, outperforming the results provided by other algorithms in the state-of-the art, in terms of the average overlap metric.

A Survey on Computer-Aided Diagnosis of Brain Disorders through MRI Based on Machine Learning and Data Mining Methodologies with an Emphasis on Alzheimer Disease Diagnosis and the Contribution of the Multimodal Fusion

“…• Self-organizing map: SOM [112,176,177] is a type of ANN that produces a discrete, low-dimensional representation of the input space for learning samples. This classifier is simple to implement and easy to understand.…”

“…However, this classifier does not exclude exponential functions, and its ability to follow trends over time is slow. • Support vector machines: SVMs used in several works [6,159,[173][174][175][176] have high generalization performance, especially when the dimension of the function space is very large. These machines offer the possibility of training generalizable nonlinear classifiers in large spaces using a small learning set.…”

Diagnosis and Classification of Brain Tumor

“…This technique provieds an accuracy of 98% sensitivity of a 100% for the mind pictures. Vaishnavee et al [25] had worked on segmentation of MRI picture the usage of SOM clustering. They had used Histogram Equalization for extraction of capabilities that improved the segmentation accuracy.…”

Genetic Algorithm Based Feature Selection Brain Tumour Segmentation and Classification

“…MRI uses magnetic field to diagnosis any changes occurring inside the brain and provide high quality results. CT scan uses radiations to check for abnormalities inside the tissues [7,8]. Normally, brain tumor affects the Cerebral Spinal Fluid (CSF), it causes for strokes.…”