A transfer learning based deep learning model to diagnose covid-19 CT scan images.

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Computational Biology

Background:

• The COVID-19 pandemic necessitates efficient diagnostic tools, especially when resources are strained.
• Accurate and rapid patient diagnosis is critical for effective treatment and resource allocation.

Purpose of the Study:

• To develop and validate an automated deep learning-based method for diagnosing COVID-19 from medical images.
• To compare the efficacy of different image segmentation techniques and deep learning models for COVID-19 detection.

Main Methods:

• CT scan images were pre-processed and segmented using various algorithms, including Otsu's method.
• Deep learning models (ResNet-50, MobileNet, VGG-16) were employed for image classification.
• Performance was evaluated based on accuracy, sensitivity, precision, and specificity.

Main Results:

• Otsu's algorithm demonstrated superior performance in image segmentation compared to other methods.
• The VGG-16 model, combined with Otsu's segmented CT images, achieved a high diagnostic accuracy of 99.28%.
• The proposed method significantly outperformed non-segmented image analysis.

Conclusions:

• Deep learning, particularly the VGG-16 model with Otsu's segmentation, offers a highly accurate and efficient solution for COVID-19 diagnosis.
• This automated approach can reduce diagnostic complexity and time, aiding in critical pandemic situations.
• Integration with Arterial Blood Gas (ABG) analysis allows for severity assessment and tailored treatment protocols.