Computational Intelligence Approach to Improve the Classification Accuracy of Brain Neoplasm in MRI Data

Abstract

Of late, in many medical diagnostic applications, automatic detection of brain neoplasm in Magnetic Resonance Imaging (MRI) data is gaining importance. This report presents two improvements for brain neoplasm detection in MRI data: 1) an advanced preprocessing technique to accurately identify the region of interest in MRI data and 2) a hybrid technique using Convolutional Neural Network (CNN) for feature extraction followed by modified Support Vector Machine (SVM) for classification. Toward the advanced preprocessing, contour calibration, Sobel edge detection and contrast optimization, which were ignored in the existing works have been investigated in this work. The autoencoder of CNN model has been used to extract the feature automatically in contrast to the manually defined feature as in the conventional techniques. The learning algorithm for SVM is modified with the addition of a cost function to minimize the false positive prediction addressing the errors in MRI data diagnosis. The proposed approach efficiently detects the presence of neoplasm and also predict whether it is malignant or benign. To check the effectiveness of the proposed preprocessing technique, it was inspected visually and then evaluated using the performance of prediction accuracy. A comparison study between the proposed classification technique and the existing techniques was performed. The result showed that the proposed approach outperformed in terms of accuracy and can handle errors in classification better than the existing approaches.

Comparative Evaluation Of Machine Learning Classifiers For Brain Tumor Detection

“…This paper [19] highlights the "curse of dimensionality" often encountered in brain tumor datasets with many features. They propose a two-pronged approach: first, using Particle Swarm Optimization to select the most informative features, mimicking the efficient foraging behavior of birds or fish.…”