Deep learning MRI radiomics nomogram predicts gross extrathyroidal extension in papillary thyroid carcinoma

In a 101-patient study, a deep learning nomogram using multiphase contrast-enhanced MRI predicted gross extrathyroidal extension with an AUC of 0.895 in validation, supporting noninvasive preoperative assessment.

• The Min-modality deep learning radiomic signature alone yielded a validation AUC of 0.853 for ETE prediction.
• In ten-fold cross-validation, the nomogram's average AUC was 0.846; the study lacked external validation as a single-center retrospective feasibility analysis.

Automated summary

RadPigeon summaries are original and for information only. They are not clinical advice.