Hybrid deep learning and feature optimization approach for early detection of multiple sclerosis

Area of Science:

• Medical Imaging
• Artificial Intelligence in Healthcare
• Neurology

Background:

• Healthcare increasingly uses autonomous systems for Multiple Sclerosis (MS) detection to reduce diagnostic delays and disability.
• Accurate and timely diagnosis of MS is crucial for effective treatment and improved patient outcomes.
• Current diagnostic methods can be resource-intensive and time-consuming.

Purpose of the Study:

• To propose a hybrid framework for accurate Multiple Sclerosis (MS) classification using deep learning, metaheuristic optimization, and machine learning.
• To enhance the efficiency and reliability of automated MS diagnosis.
• To evaluate the performance of different machine learning classifiers with optimized features.

Main Methods:

• MRI images were preprocessed using CLAHE, resizing, and normalization.
• Deep features were extracted using a pretrained VGG16 convolutional neural network (CNN).
• The Whale Optimization Algorithm (WOA) was used for feature selection, optimizing for Support Vector Machine (SVM) performance, followed by evaluation of multiple classifiers including Artificial Neural Network (ANN).

Main Results:

• The hybrid framework successfully extracted deep features and reduced dimensionality using WOA.
• The Artificial Neural Network integrated with WOA (ANN+WOA) achieved the highest classification accuracy of 98%.
• The proposed model demonstrated high performance in automated MS diagnosis.

Conclusions:

• The developed hybrid framework shows significant potential for reliable, efficient, and automated Multiple Sclerosis (MS) diagnosis.
• Integrating deep learning with metaheuristic feature selection can substantially improve classification accuracy.
• This approach offers a promising tool to aid clinicians in MS diagnosis and management.