Metaheuristic optimizers integrated with vision transformer model for severity detection and classification via...

Area of Science:

• Artificial Intelligence
• Medical Imaging
• Computational Biology

Background:

• Accurate COVID-19 severity classification is crucial for effective patient management and resource allocation.
• Traditional methods for analyzing medical images like X-rays and CT scans can be time-consuming and subjective.
• The need for automated, efficient, and accurate diagnostic tools is paramount in managing large-scale health crises.

Purpose of the Study:

• To develop and evaluate a novel hybrid framework for classifying COVID-19 severity using chest X-rays (CXR) and computed tomography (CT) scans.
• To integrate Vision Transformers (ViT) with metaheuristic optimization techniques for enhanced feature extraction and classification accuracy.
• To compare the performance of the proposed framework against traditional Convolutional Neural Network (CNN) based approaches.

Main Methods:

• A hybrid framework combining Vision Transformers (ViT) with Grey Wolf Optimizer (GWO) for hyperparameter tuning and Particle Swarm Optimization (PSO) for feature selection.
• Utilizing ViT's self-attention mechanism to extract global and local image features from CXR and CT scans.
• Implementing a multi-phase classification strategy to distinguish between normal, mild, moderate, and severe COVID-19 cases.

Main Results:

• The GWO_ViT_PSO_MLP model achieved high accuracy: 99.14% for 2-class CXR classification and 98.89% for 2-class CT classification.
• Outperformed traditional CNN models like ResNet34 (84.22%) and VGG19 (93.24%) in binary classification tasks.
• Demonstrated superior performance in multi-class severity classification, particularly in differentiating mild and moderate COVID-19 cases.

Conclusions:

• The proposed hybrid framework offers a significant improvement in accuracy and computational efficiency for COVID-19 severity classification.
• The model shows potential as a scalable and reliable tool for automated detection and severity assessment in clinical settings.
• This approach advances the application of AI in medical imaging for infectious disease management.