A deep learning approach using effective preprocessing techniques to detect COVID-19 from chest CT-scan and X-ray...

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Infectious Diseases

Background:

• Coronavirus disease 2019 (COVID-19) poses a global health challenge, particularly impacting resource-limited countries with insufficient diagnostic capabilities.
• Limited availability of COVID-19 diagnostic kits necessitates innovative approaches for rapid case identification.

Purpose of the Study:

• To develop and evaluate a deep learning model for the accurate detection of COVID-19 using chest imaging.
• To provide a diagnostic support tool for clinicians and radiologists, especially in areas with limited resources.

Main Methods:

• A modified ResNet50V2 deep learning architecture was utilized.
• The model was trained on a diverse dataset of chest X-ray and CT scan images.
• Image preprocessing involved the application of a sharpening filter.

Main Results:

• High accuracy achieved: 96.45% for four-class classification (COVID-19/Normal/Bacterial/Viral Pneumonia) on X-rays.
• Exceptional accuracy on CT scans: 99.99% for Normal/COVID-19 classification and 99.01% for COVID-19/Normal/Community-acquired Pneumonia.
• The model demonstrated robust performance across various classification tasks.

Conclusions:

• Deep learning models can effectively diagnose COVID-19 from basic chest imaging, offering a valuable tool for underserved regions.
• The developed AI model shows potential to significantly enhance the speed and accuracy of COVID-19 diagnosis.
• This approach leverages widely available equipment to support global public health efforts against COVID-19.