An efficient glioma classification and grade detection using hybrid convolutional neural network-based SVM model

ABSTRACT

Glioma develops in the brain and spinal cord. Oncologists frequently use ‘low-grade’ and ‘high-grade’ to describe how quickly malignant gliomas spread. Low-grade gliomas grow slowly, but still, they are malignant, and if left untreated, they progress into high-grade gliomas. Most of the existing diagnosis approaches use traditional machine learning (ML) approaches to perform this task. However, all failed to accurately detect the early stages with a maximum accuracy rate. Technological advances and deep learning (DL) techniques are enabling radiologists to detect tumors without invasive procedures. DL models play an important role in increasing the performance of image classification tasks related to the medical field. Therefore, a novel Convolutional neural network-based Support vector machine (CNN-SVM) is proposed in this research to enhance glioma grade detection accurately in this research. The detection process contains two phases: Glioma classification and Glioma grade detection. In the first phase, the CNN classifies the glioma images. The glioma region is segmented using the modified firefly optimizer functions to extract the shape and texture information of the affected glioma regions. These extracted glioma features are trained with SVM Classifier to detect the glioma grades. This hybrid deep model-based glioma detection approach sees the grades of gliomas efficiently. The performance of this glioma detection approach is analyzed with existing models. The validation outputs show that the hybrid model has obtained a maximum accuracy rate of 99.98%. It proves that the efficiency of the hybrid model is improved Glioma grade detection accuracy rate more effectively than comparison approaches.

KEYWORDS:

• Convolution neural network
• glioma classification
• glioma grade detection
• support vector machine
• machine learning
• deep learning
• pre-processing
• feature extraction

Acknowledgment

We thank the anonymous referees for their useful suggestions.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Notes on contributors

S. Shargunam

Mr. S. Shargunam received his M.E. Degree from the Department of Computer Science and Engineering at PSG College of College of Technology, Anna University, Coimbatore, India, in 2020. He is a Research Scholar at Francis Xavier Engineering College, Tirunelveli, India. He is pursuing his Ph.D degree in Information and Communication Engineering from Anna University. His current Research interest includes Medical Image Processing, Deep Learning and Machine Learning.

G. Rajakumar

Dr. G. Rajakumar received his Ph.D degree from the Department of Information and Communication Engineering at M. S University, Tirunelveli, India, in 2014. He is a Professor in the Department of Electronics and Communication Engineering at Francis Xavier Engineering College, Tirunelveli, India. His Research interest includes Digital Image Processing, Medical Image Processing, High Speed Networks, VLSI Design, Digital Signal Processing, Control Systems, Wireless Networks and Real Time Embedded Systems.