Radiomics-based diagnosis of acute cholangitis using non-contrast computed tomography: a preliminary study

Area of Science:

• Radiology
• Medical Imaging
• Artificial Intelligence in Medicine

Background:

• Acute cholangitis diagnosis often relies on contrast-enhanced CT (CECT), which carries risks for some patients.
• Non-contrast CT (NCCT) is a safer alternative but may offer limited diagnostic information.
• Radiomics, extracting quantitative features from medical images, presents an opportunity to enhance diagnostic capabilities of NCCT.

Purpose of the Study:

• To develop and evaluate a radiomics model for diagnosing acute cholangitis using only NCCT.
• To compare the diagnostic performance of the NCCT-based radiomics model against radiologist interpretations of NCCT and CECT.

Main Methods:

• A total of 158 patients (79 with acute cholangitis, 79 controls) underwent NCCT and CECT.
• Radiomics features were extracted from segmented livers on NCCT images.
• A radiomics model was developed and validated on separate datasets, and its performance was compared to radiologists' interpretations using both NCCT and CECT.

Main Results:

• The NCCT-based radiomics model achieved an area under the receiver operating characteristic curve (AUROC) of 0.888, with 95% sensitivity, 75% specificity, and 85% accuracy in the test dataset.
• Radiologists' interpretation of NCCT alone had 65% sensitivity, 100% specificity, and 82.5% accuracy.
• Radiologists' performance improved to 95% sensitivity, 100% specificity, and 97.5% accuracy when using CECT.

Conclusions:

• NCCT-based radiomics models show significant potential for diagnosing acute cholangitis.
• This AI-driven approach offers a viable alternative to CECT, especially for patients with contraindications to contrast agents.
• Further validation is warranted to integrate this technique into clinical practice for acute cholangitis diagnosis.