COVID-19 Case Recognition from Chest CT Images by Deep Learning, Entropy-Controlled Firefly Optimization, and...

Area of Science:

• Medical imaging analysis
• Artificial intelligence in healthcare
• Deep learning for disease diagnosis

Background:

• Healthcare generates vast data from medical devices like CT scanners.
• Artificial intelligence (AI) methods analyze this data for early disease detection.
• Deep learning, particularly computer vision, is crucial for diagnosing COVID-19 from medical images.

Purpose of the Study:

• To propose a novel automated technique for COVID-19 diagnosis using AI.
• To enhance the accuracy of COVID-19 classification in medical images.
• To integrate parallel fusion and optimization of deep learning models for improved diagnostic performance.

Main Methods:

• Image preprocessing included contrast enhancement using top-hat and Wiener filters.
• Two pre-trained deep learning models, AlexNet and VGG16, were fine-tuned for COVID-19 and healthy class classification.
• Features were extracted, fused using parallel positive correlation, optimized via entropy-controlled firefly optimization, and classified using multiclass support vector machine (MC-SVM).

Main Results:

• The proposed automated technique achieved a high accuracy of 98% in classifying COVID-19 cases.
• The method demonstrated improved performance compared to existing schemes through detailed analysis.
• The parallel fusion and optimization approach effectively enhanced diagnostic capabilities.

Conclusions:

• The developed AI-driven technique offers a promising automated solution for accurate COVID-19 diagnosis from medical images.
• The integration of advanced deep learning, feature fusion, and optimization techniques significantly improves diagnostic accuracy.
• This approach supports medical professionals in early and reliable identification of COVID-19 patients.