COV-DLS: Prediction of COVID-19 from X-Rays Using Enhanced Deep Transfer Learning Techniques.

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Computer Science

Background:

• Accurate and rapid diagnosis of COVID-19 is crucial for effective patient management and public health.
• Chest X-rays are a widely accessible imaging modality for detecting pneumonia and related lung conditions.

Purpose of the Study:

• To develop and evaluate a novel deep learning system (COV-DLS) for accurate COVID-19 classification using chest X-ray images.
• To enhance existing deep learning architectures (VGG16, VGG19, ResNet50, InceptionV3) for improved diagnostic performance.

Main Methods:

• Modification of VGG16, VGG19, ResNet50, and InceptionV3 architectures by incorporating average pooling, dense layers, and dropout layers.
• Implementation of a two-phase COV-DLS system: heading model construction and classification.
• Application of the modified architectures on a COVID-19 chest X-ray image dataset.

Main Results:

• Modified-VGG16 achieved 98.61% accuracy, Modified-VGG19 achieved 97.22%, Modified-ResNet50 achieved 95.13%, and Modified-InceptionV3 achieved 99.31%.
• The proposed COV-DLS system demonstrated superior performance compared to existing deep learning models in terms of classification accuracy and F1-score.

Conclusions:

• The modified deep learning architectures, particularly Modified-VGG16 and Modified-InceptionV3 within the COV-DLS framework, show significant promise for automated COVID-19 detection.
• COV-DLS offers a robust and accurate solution for COVID-19 classification from chest X-rays, potentially aiding clinical diagnosis.