Brain tumor classification based on deep CNN and modified butterfly optimization algorithm

ABSTRACT

There is an emerging need for medical imaging data to provide timely diagnosis. The segmentation of brain tumours from Magnetic Resonance Imaging (MRI) is of great significance for planning treatment. However, mechanising the process with different imaging conditions and accuracy is a significant challenge due to variations in tumour structures. An effective optimisation-driven classifier is developed for brain tumour detection to increase accuracy. The proposed Modified Butterfly Optimization Algorithm (MBOA) is employed for training the Deep Convolutional Neural Network (DCNN) for brain tumour detection. The input MRI image is pre-processed using a Gaussian filter to eliminate noises. The pre-processed output is fed to segmentation, wherein the U-Net model is adapted for generating the segments. The extraction of statistical and texture features is done for tumour region classification. The classification of tumours is done with DCNN, wherein the weights are optimally tuned using the proposed MBOA. The experimental result demonstrates that the developed method achieved better.

KEYWORDS:

• Tumor detection
• segmentation
• LGBP
• optimisation
• DCNN

Disclosure statement

No potential conflict of interest was reported by the authors.