Feature fusion based VGGFusionNet model to detect COVID-19 patients utilizing computed tomography scan images

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Computational Biology

Background:

• The COVID-19 pandemic poses a significant global health threat, necessitating rapid and accurate diagnostic tools.
• Early detection of COVID-19 is crucial for effective patient management and disease control.
• Computed Tomography (CT) scans offer valuable insights for diagnosing respiratory illnesses, including COVID-19.

Purpose of the Study:

• To develop and evaluate an automated deep learning approach for detecting COVID-19 from CT scan images.
• To compare the performance of various pre-trained Convolutional Neural Network (CNN) models for COVID-19 detection.
• To establish a reliable AI-assisted tool for augmenting clinical diagnosis of COVID-19.

Main Methods:

• A four-phase paradigm was implemented: image preprocessing, noise reduction using anisotropic diffusion, image segmentation, and CNN model training/testing.
• Well-established pre-trained CNN models (AlexNet, ResNet50, VGG16, VGG19) were utilized.
• A dataset of 4861 real-life COVID-19 CT images (3068 positive, 1793 negative) was used, with 80% for training and 20% for testing.

Main Results:

• The VGG19 pre-trained CNN model demonstrated superior performance, achieving an accuracy of 98.06%.
• The proposed deep learning approach showed high accuracy in distinguishing between COVID-19 positive and negative CT scans.
• Experimental results confirm the efficacy of the automated detection system.

Conclusions:

• The developed automated deep learning method, particularly using the VGG19 model, is highly accurate for COVID-19 detection from CT scans.
• This AI-driven approach can serve as a valuable assistant to healthcare professionals for timely and accurate diagnosis.
• The study highlights the potential of deep learning in enhancing diagnostic capabilities during infectious disease outbreaks.