Creative Commons source reading · Machine learning for performance
Hybrid deep learning and feature optimization approach for early detection of multiple sclerosis
Nandini Anam; Sharief Basha S.; Chiranji Lal Chowdhary
Originally published in Front Hum Neurosci · 2026-01-12
From the article
Selected source text. JournalOS removes markup and shortens excerpts where indicated. The full article provides the complete context.
Abstract
The healthcare field increasingly relies on autonomous systems for the detection and analysis of Multiple Sclerosis (MS) to minimize diagnostic delays, resource burdens, reduce the progression of disability, and enhance clinical decision-making efficiency. Such systems ensure accurate and timely treatment, ultimately improved patient outcomes. In this study, a hybrid framework combining deep learning-based feature extraction, metaheuristic feature selection, and machine learning (ML) classifiers is proposed for accurate MS classification. All MRI images were preprocessed using Contrast Limited Adaptive Histogram Equalization (CLAHE), resizing, and normalization to enhance contrast and standardize the input dimensions. Deep features were extracted using the pretrained VGG16 convolutional neural network (CNN), in which the fully…
Results and discussion
To evaluate the effectiveness of the proposed method, we used a dataset containing 3,427 brain MRI images related to Multiple Sclerosis, collected from Kaggle. The dataset is categorized into four classes. The pre-processing was applied to the images to ensure uniformity in size and contrast. Data was split into 80% training and 20% testing sets to evaluate model performance. The training set was used for feature extraction and model learning, while the test set was used to assess classification accuracy. CLAHE was first used to enhance image contrast, followed by feature extraction using the CNN (VGG16) model. The most relevant features were then selected using the Whale Optimization Algorithm. In this study, feature extraction was performed on preprocessed MRI images without segmentation. A pretrained VGG16 model was used to extract…
Conclusion
To improve the accuracy of MS detection and support early clinical intervention, this study addresses critical challenges in medical imaging such as computational efficiency, generalization, and effective feature extraction through a robust hybrid classification framework. The proposed system integrates CLAHE-based contrast enhancement, deep feature learning using VGG16, and bio-inspired Whale Optimization Algorithm for optimal feature selection. This approach significantly improves the model’s ability to identify disease-related patterns in brain MRI scans. The model demonstrated exceptional performance, particularly the ANN+WOA classifier, which achieved a classification accuracy of 98%, outperforming both SVM and RF classifiers. The WOA-based feature selection effectively reduced dimensionality while retaining discriminative…
Editorial reading guide
- What data created and tested the model, and was evaluation independent?
- What is the cost of a false positive, false negative or out-of-distribution decision?
