An automatic radiomic-based approach for disease localization: A pilot study on COVID-19.

Area of Science:

• Radiomics and Medical Imaging
• Personalized Medicine
• Artificial Intelligence in Healthcare

Background:

• Radiomics facilitates diagnosis and prognosis by extracting quantitative features from medical images.
• Current radiomics methods often rely on manual or semi-automated Region of Interest (ROI) definition, which can be time-consuming and prone to errors.
• Minimizing diagnostic errors, such as false positives and false negatives, is crucial for effective patient care.

Purpose of the Study:

• To develop an automated approach for detecting and localizing disease-specific areas in medical images.
• To minimize diagnostic errors by refining the feature extraction process in radiomics.
• To evaluate the proposed matrix-based radiomics method using a case study of COVID-19.

Main Methods:

• A novel radiomics approach creating an nxn grid on DICOM image sequences to extract features from individual cells.
• Utilizing Model Checking techniques for automated analysis and diagnosis.
• Comparing the proposed grid-based method against traditional single-ROI extraction techniques.

Main Results:

• The proposed method successfully identifies and localizes disease markers, demonstrating promising results in the COVID-19 case study.
• The grid-based radiomics approach significantly improved diagnostic accuracy and recall compared to methods using the entire image as a single ROI.
• The approach enhances knowledge, interoperability, and trust in radiomics software tools.

Conclusions:

• The developed matrix-based radiomics method offers a more accurate and localized approach to disease diagnosis.
• This automated technique has the potential to reduce diagnostic errors and improve patient outcomes.
• The study supports the advancement of collaborative and trustworthy AI-driven tools in medical diagnostics.