Enhanced Disc Herniation Classification Using Grey Wolf Optimization Based on Hybrid Feature Extraction and Deep...

Area of Science:

• Medical Imaging Analysis
• Artificial Intelligence in Healthcare
• Spinal Diagnostics

Background:

• Increasing incidence of lumbar disc herniation due to prolonged computer use.
• Early diagnosis and treatment are crucial for favorable outcomes.
• Need for automated, accurate classification of lumbar disc herniations.

Purpose of the Study:

• To develop a computer-aided, fully automated method for classifying lumbar disc herniations using MRI.
• To enhance diagnostic accuracy and efficiency through advanced machine learning techniques.

Main Methods:

• Hybrid approach combining Residual Network (ResNet50) for feature extraction and Grey Wolf Optimization (GWO) for feature selection.
• Utilized Multi-Layer Perceptron (MLP) and Support Vector Machine (SVM) classifiers with various functions for classification.
• Employed deep learning for robust feature representation and optimization algorithms for enhanced classification performance.

Main Results:

• The proposed ResNet50-GWO-SVM methodology demonstrated significant improvements in accuracy, precision, recall, and F1 score.
• Outperformed traditional approaches in classifying lumbar disc herniations.
• Validated the effectiveness of integrating deep learning feature extraction with optimization and machine learning classifiers.

Conclusions:

• The ResNet50-GWO-SVM approach achieved superior performance compared to CapsNet, EfficientNetB6, and DenseNet169.
• Demonstrated robustness and effectiveness in lumbar disc herniation classification tasks.
• Highlights the potential of AI-driven tools for improved spinal diagnostics.