Non-contrast CT-based radiomic signature for screening thoracic aortic dissections: a multicenter study

Area of Science:

• Radiology
• Medical Imaging
• Artificial Intelligence in Medicine

Background:

• Thoracic aortic dissection (AD) is a life-threatening condition requiring timely diagnosis.
• Current diagnostic methods may involve contrast-enhanced imaging, posing risks for certain patients.
• Developing non-contrast CT-based screening tools is crucial for improving AD detection efficiency.

Purpose of the Study:

• To develop and validate a radiomic signature derived from non-contrast CT scans for screening thoracic aortic dissections (ADs).
• To assess the diagnostic performance of this radiomic signature in a multi-center cohort.
• To explore the potential of this non-contrast CT-based approach as a screening tool for AD.

Main Methods:

• Retrospective analysis of 378 patients with non-contrast chest CT scans from four medical centers.
• Extraction of radiomic features from non-contrast CT images.
• Development of a radiomic signature using logistic regression, validated across training, validation, and external test sets.

Main Results:

• The radiomic signature achieved high predictive performance with Area Under the Curve (AUC) values of 0.91, 0.92, and 0.90 in the training, validation, and external test sets, respectively.
• In the external test group, the signature demonstrated 90.5% diagnostic accuracy, 85.7% sensitivity, and 91.7% specificity.
• The radiomic signature showed superior predictive performance compared to existing clinical models.

Conclusions:

• A non-contrast CT-based radiomic signature is a highly effective tool for predicting thoracic aortic dissections (ADs).
• This radiomic signature offers a promising, non-invasive screening method for AD.
• The developed signature has the potential to be integrated into clinical practice for early AD detection.