Multiple Sclerosis Identification Based on Fractional Fourier Entropy and a Modified Jaya Algorithm.

Area of Science:

• Medical Imaging Analysis
• Artificial Intelligence in Healthcare
• Neurology

Background:

• Human identification of multiple sclerosis (MS) faces challenges with normal-appearing white matter, leading to reduced sensitivity.
• Automated methods are needed to improve the accuracy and efficiency of MS diagnosis.

Purpose of the Study:

• To develop and validate a computer vision-based approach for the automatic identification of multiple sclerosis (MS).
• To enhance diagnostic sensitivity and accuracy compared to traditional methods.

Main Methods:

• Extraction of fractional Fourier entropy maps from brain images.
• Feature classification using a multilayer perceptron trained with an improved parameter-free Jaya algorithm.
• Implementation of cost-sensitivity learning to address imbalanced data.

Main Results:

• Achieved high diagnostic performance with 97.40 ± 0.60% sensitivity, 97.39 ± 0.65% specificity, and 97.39 ± 0.59% accuracy via 10 × 10-fold cross-validation.
• The improved Jaya algorithm demonstrated superior classification performance and training speed compared to the standard Jaya algorithm and other bioinspired methods.
• The proposed method outperformed four state-of-the-art MS identification approaches.

Conclusions:

• The developed computer vision approach offers a robust and accurate method for automated MS identification.
• The improved Jaya algorithm provides significant advantages in training speed and classification accuracy for medical image analysis.
• This method represents a substantial advancement over existing techniques for diagnosing multiple sclerosis.