Hybrid RGSA and Support Vector Machine Framework for Three-Dimensional Magnetic Resonance Brain Tumor Classification

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Neuroscience

Background:

• Medical image classification is crucial for clinical diagnosis and research.
• Magnetic resonance imaging (MRI) is a key modality for detecting brain abnormalities.
• Accurate brain tumor classification is essential for effective treatment planning.

Purpose of the Study:

• To propose a novel three-dimensional (3D) hybrid model for brain tumor classification using MRI.
• To differentiate between benign and malignant brain tumors based on micro- and macroscale textures.
• To enhance the accuracy of brain tumor identification through advanced feature extraction and selection.

Main Methods:

• Preprocessing of MRI images using a 3D Gaussian filter.
• Feature extraction via 3D volumetric Square Centroid Lines Gray Level Distribution Method (SCLGM) and 3D texture matrices.
• Optimal feature selection using a refined gravitational search algorithm (RGSA).
• Classification using Support Vector Machines, Backpropagation Network, and K-Nearest Neighbor algorithms.
• System evaluation using a leave-one-case-out method on 320 real-time brain MRI images.

Main Results:

• The proposed RGSA effectively selected optimal features for classification.
• The hybrid model achieved a high classification performance, indicated by a receiver operating characteristic curve of 0.986 (±0.002).
• Experimental results demonstrated the efficiency of the feature extraction and selection algorithms.

Conclusions:

• The developed 3D hybrid approach offers a systematic and efficient method for brain tumor classification from MRI.
• The study highlights the potential of combining advanced texture analysis with intelligent optimization algorithms for improved diagnostic accuracy.
• This model shows promise for clinical application in differentiating brain tumor types.