Diagnosis of Covid-19 from CT slices using Whale Optimization Algorithm, Support Vector Machine and Multi-Layer...

Area of Science:

• Medical Imaging
• Artificial Intelligence in Healthcare
• Diagnostic Systems

Background:

• Coronavirus disease 2019 (Covid-19) is a severe respiratory illness caused by SARS-CoV-2.
• Accurate and timely diagnosis is crucial for managing Covid-19.
• Chest Computed Tomography (CT) is a key imaging modality for Covid-19 detection.

Purpose of the Study:

• To develop and evaluate a Computer Aided Diagnosis (CAD) system for Covid-19 detection using chest CT slices.
• To enhance diagnostic accuracy through optimized feature selection.
• To compare the proposed system's performance against established machine learning classifiers.

Main Methods:

• Lung tissue segmentation using Otsu's thresholding method.
• Identification and annotation of Covid-19 lesions as Regions of Interest (ROIs).
• Feature extraction (texture and shape), selection via Whale Optimization Algorithm (WOA) and Support Vector Machine (SVM) accuracy, and classification using Multi-Layer Perceptron (MLP).

Main Results:

• The proposed CAD system achieved 88.94% accuracy with feature selection.
• The MLP classifier without feature selection yielded 80.40% accuracy.
• The system significantly outperformed eight benchmark Machine Learning classifiers on a real-time dataset.

Conclusions:

• Feature selection using WOA substantially improves the diagnostic accuracy of the MLP classifier for Covid-19 detection.
• The developed CAD system demonstrates high efficacy and potential for clinical application in diagnosing Covid-19 from chest CT scans.
• Statistical analyses confirm the significant impact and superiority of the proposed method on the considered dataset.