Radiomics Based on Magnetic Resonance Imaging in Differentiation of Hepatic micro-abscess from Hepatic micro-metastasis: A Multicenter Study

Motivation: The clinical and imaging characteristics of hepatic micro-abscesses and hepatic micro-metastases may exhibit certain similarities. Goal(s): To evaluate the feasibility of using magnetic resonance imaging (MRI)-based radiomics model to differentiate hepatic micro-abscess from hepatic micro-metastasis. Approach: MRI scans using 1.5 or 3.0 T systems from various manufacturers and the uAI Research Portal (United Imaging Intelligence, China) were used for image acquisition and post-processing. Chi-square test, Binary logistic regression analysis, and Delong test were used for statistical analysis. Results: Various machine learning models utilizing semantic features demonstrate high efficacy in distinguishing hepatic micro-abscesses from hepatic micro-metastases. Impact: Our findings regarding Differentiation of Hepatic micro-abscess from Hepatic micro-metastasis is essential in selecting the best treatment option and enhancing prognosis.