COVID-19 detection from lung CT-Scans using a fuzzy integral-based CNN ensemble

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Computational Biology

Background:

• The COVID-19 pandemic severely impacted global healthcare systems and economies.
• Limited population screening and slow, less sensitive RT-PCR tests hindered effective SARS-CoV-2 containment.
• The need for rapid, accurate, and accessible COVID-19 detection methods is critical.

Purpose of the Study:

• To develop a highly accurate and sensitive computer-aided detection (CAD) system for COVID-19 using chest CT scans.
• To leverage deep learning and ensemble methods for improved classification of COVID-19 from non-COVID-19 cases.
• To address the limitations of existing COVID-19 diagnostic tools.

Main Methods:

• Proposed a Sugeno fuzzy integral ensemble of four pre-trained deep learning models: VGG-11, GoogLeNet, SqueezeNet v1.1, and Wide ResNet-50-2.
• Utilized chest CT-scan images for classification into COVID-19 and Non-COVID-19 categories.
• Evaluated the framework on a publicly available dataset.

Main Results:

• Achieved 98.93% accuracy and 98.93% sensitivity in classifying COVID-19 cases.
• The proposed model demonstrated superior performance compared to state-of-the-art methods on the same dataset.
• The ensemble deep learning approach proved to be a reliable COVID-19 diagnostic tool.

Conclusions:

• The developed deep learning-based CAD system offers a promising and effective solution for COVID-19 detection.
• This method provides a sensitive and accurate alternative to traditional diagnostic techniques.
• The framework's high performance suggests its potential for widespread clinical application in pandemic scenarios.