Objective:

To develop a deep-learning-based radiomics nomogram (DLRN) that integrates visceral adipose features from non-contrast CT for preoperative risk stratification of gastrointestinal stromal tumors (GISTs).

Approach:

Key Findings:

• The DLRN achieved an AUC of 0.936 in the derivation cohort and 0.862 in the external test cohort.
• Accuracy was 0.873 in the derivation cohort and 0.925 in the external test cohort.
• Sensitivity and specificity were 0.811 and 0.904 in the derivation cohort, and 0.833 and 0.936 in the external test cohort, respectively.
• The DLRN provided a higher net clinical benefit than all comparison models in both datasets.

Interpretation:

The DLRN may serve as a non-invasive tool for preoperative GIST risk stratification using non-contrast CT.

Limitations:

• Further validation in larger multicentre cohorts is needed.

Conclusion:

The DLRN integrates visceral fat-derived radiomics and clinical variables for GIST risk stratification.

Attribution Notice

This content is an AI-generated, fully rewritten summary based on a published scholarly article. It does not reproduce the original text and is not a substitute for the original publication. Readers are encouraged to consult the source for full context, data, and methodology.