A MACHINE LEARNING APPROACH TO BRAIN TUMORS SEGMENTATION USING ADAPTIVE RANDOM FOREST ALGORITHM

Abstract

In this paper a brain tumor segmentation method is proposed which is based on the Random Forest algorithm. The proposed technique is applied to the brain magnetic resonance images and the performance indices including Dice Similarity Coefficient (DSC) as well as algorithm accuracy (ACC) are calculated that are 98.38% and 97.65%, respectively. The obtained results show that the proposed model can have a good performance when compared with the other segmentation methods. Besides, in this paper the mathematical modeling of the Random Forest algorithm is provided.

Existing System

Cohesion-based self-merging (CSM) algorithm, Support Vector Machine (SVM)

Proposed System

This method proposed a randomized segregation model based on Random Forest (RF) algorithm for the segmentation of the brain tumors in 3D MR images. During the formation of RF classifier, this algorithm evaluates the importance of variables, the relationship between MRI data and lastly generalized error. These three features of the RF are presented to optimize the partitioning framework. The proposed model in this paper has been tested on BraTS 3D MRI datasets and the obtained results together with the dice coefficient show improvement as compared to recent methods.

Architecture

BLOCK DIAGRAM