COVID-CCD-Net: COVID-19 and colon cancer diagnosis system with optimized CNN hyperparameters using gradient-based...

Area of Science:

• Medical Imaging Analysis
• Artificial Intelligence in Healthcare
• Computational Pathology

Background:

• Coronavirus disease-2019 (COVID-19) pandemic necessitates rapid diagnostics, where RT-PCR kits are limited.
• X-ray imaging offers accessible COVID-19 detection, outperforming limited PCR tests.
• Colorectal cancer diagnosis relies on Tissue Microarray (TMA) analysis, requiring advanced computational methods.

Purpose of the Study:

• To develop an optimized deep learning model for classifying COVID-19 from chest X-rays and colorectal cancer subtypes from TMAs.
• To enhance the accuracy and efficiency of medical image classification using artificial intelligence.

Main Methods:

• Proposed COVID-CCD-Net, a convolutional neural network (CNN) classification approach.
• Utilized Gradient-based Optimizer (GBO) algorithm for hyperparameter optimization.
• Integrated various CNN architectures (AlexNet, DarkNet-19, Inception-v3, MobileNet, ResNet-18, ShuffleNet) and tested on COVID-19 and Epistroma datasets.

Main Results:

• Achieved accurate classification of COVID-19, normal, and viral pneumonia in chest X-rays.
• Successfully classified epithelial and stromal regions in colorectal cancer TMAs.
• Demonstrated significant improvement in classification performance compared to non-optimized CNNs, even at low epochs.

Conclusions:

• The COVID-CCD-Net model offers a highly effective solution for medical image classification tasks.
• Optimized CNNs show superior performance in diagnosing infectious diseases and cancers.
• This approach holds promise for improving early disease detection and patient outcomes.