CNGOD-An improved convolution neural network with grasshopper optimization for detection of COVID-19

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Computational Biology

Background:

• The COVID-19 pandemic, caused by SARS-CoV-2, necessitates rapid and accessible diagnostic methods.
• Current laboratory-based tests for COVID-19 require specialized kits and can be time-consuming.
• Chest X-rays offer a widely available alternative for disease diagnosis, but automated analysis requires advanced techniques.

Purpose of the Study:

• To develop an automated diagnostic method for COVID-19 using chest X-ray images.
• To propose a novel deep convolution neural network (CNN) optimized with the Grasshopper Optimization Algorithm (GOA) for enhanced diagnostic accuracy.
• To address the limitations of existing diagnostic methods by leveraging readily available radiological data.

Main Methods:

• A deep CNN architecture featuring depthwise separable convolutions for efficient feature extraction.
• Optimization of the CNN model using the Grasshopper Optimization Algorithm (GOA), a metaheuristic approach known for fast convergence and effective search.
• Implementation of the Maximum Probability Based Cross Entropy Loss (MPCE) function to minimize backpropagation errors and improve training.
• Utilizing Grad-CAM for class activation mapping to interpret model predictions and enhance result transparency.

Main Results:

• The proposed GOA-optimized CNN model demonstrated high classification accuracy in identifying COVID-19 from chest X-ray images.
• The method effectively overcomes the limitations associated with traditional diagnostic kits.
• Visualizations using Grad-CAM provided interpretability for the model's diagnostic decisions.

Conclusions:

• The developed deep learning model offers a promising, automated approach for COVID-19 diagnosis using chest X-rays.
• This AI-driven method can significantly aid in rapid disease detection, especially in resource-limited settings.
• The integration of GOA for optimization enhances the efficiency and accuracy of AI-based medical image analysis for infectious diseases.