Novel Coronavirus and Common Pneumonia Detection from CT Scans Using Deep Learning-Based Extracted Features.

Area of Science:

• Medical Imaging Analysis
• Artificial Intelligence in Healthcare
• Infectious Disease Diagnostics

Background:

• COVID-19 remains a global health challenge, with current vaccines reducing severity but not preventing infection.
• Continuous testing is necessary, but manual monitoring is time-consuming and challenging due to symptom overlap with other respiratory illnesses.
• Existing over-the-counter tests are unreliable, leading to unnecessary hospital visits and further diagnostic burdens.

Purpose of the Study:

• To develop an automated system for accurate COVID-19 detection and diagnosis from chest CT scans.
• To address the urgent need for reliable diagnostic tools applicable to current and future pandemics.
• To leverage machine learning and deep learning for enhanced diagnostic capabilities without human intervention.

Main Methods:

• A modified machine learning (ML) process integrating deep learning (DL) algorithms for feature extraction.
• Utilized GoogleNet and ResNet18 for extracting 2000 features from chest CT scans.
• Employed the support vector machine (SVM) classifier for accurate COVID-19 detection.

Main Results:

• Achieved a highest average accuracy of 99.9% in detecting COVID-19 from chest CT scans.
• The modified ML process demonstrated superior performance compared to existing literature using similar datasets.
• The developed system shows significant added value to the current body of knowledge in diagnostic AI.

Conclusions:

• The proposed ML-DL integrated system provides a highly accurate and automated method for COVID-19 diagnosis.
• This technology holds potential for application in hospitals and can enhance preparedness for future health crises.
• Further research is needed to implement and validate these methods in clinical settings.