An optimized features selection approach based on Manta Ray Foraging Optimization (MRFO) method for parasite malaria...

Area of Science:

• Medical diagnostics
• Computer vision
• Machine learning

Background:

• Malaria remains a significant global health threat, causing millions of cases and deaths annually.
• Current malaria diagnosis relies heavily on manual microscopy, which is labor-intensive and requires expert interpretation.
• There is a critical need for automated, accurate, and efficient malaria detection systems.

Purpose of the Study:

• To develop and evaluate a novel automated system for malaria cell segmentation and classification.
• To enhance the accuracy and efficiency of malaria diagnosis through computational methods.

Main Methods:

• Malaria cells were segmented using a color-based k-mean clustering algorithm.
• Deep features were extracted using pre-trained models (efficient-net-b0, shuffle-net) and optimized with the Manta-Ray Foraging Optimization (MRFO) method.
• Classification was performed using Support Vector Machine (SVM) with a linear kernel, employing 10-fold cross-validation.

Main Results:

• The proposed automated system achieved a high classification accuracy of 99.2%.
• Fusion of features extracted from both efficient-net-b0 and shuffle-net models yielded superior performance compared to individual feature sets.
• The developed method outperformed existing state-of-the-art approaches in malaria detection.

Conclusions:

• The automated malaria diagnosis system demonstrates high accuracy and effectiveness.
• Feature fusion and optimization techniques significantly improve classification performance.
• This computational approach offers a promising alternative to traditional microscopy for malaria diagnosis.