Preoperative prediction of extrathyroidal extension in papillary thyroid carcinoma: a nomogram integrating dual-plane ultrasound-based radiomics and clinical features
Source: PubMed Central Open Access, NCBI / U.S. National Library of Medicine
Background Extrathyroidal extension (ETE) is an important prognostic factor in papillary thyroid carcinoma (PTC) that directly influences surgical decision-making, yet its accurate preoperative assessment by conventional ultrasound remains challenging. This study aimed to establish a nomogram integrating dual-plane (transverse and longitudinal combined) ultrasound radiomics and clinical features for the preoperative prediction of ETE in PTC. Methods A retrospective study was conducted on 483 patients with surgically confirmed PTC at our institution between January 2020 and December 2023 as the training cohort, and an additional 103 patients treated between January 2024 and June 2024 as the independent validation cohort. Radiomics features were extracted from transverse and longitudinal plane ultrasound images. The least absolute shrinkage and selection operator (LASSO) with five-fold cross-validation was applied for feature selection. Independent clinical predictors were identified by univariate and multivariate logistic regression analyses. A model integrating radiomics and clinical predictors for predicting ETE in PTC was constructed and presented as a nomogram. Results The combined model exhibited superior predictive performance, with areas under the receiver operating characteristic (ROC) curve (AUCs) of 0.880 in the training cohort and 0.858 in the independent validation cohort, and significantly outperformed both the clinical model and radiomics model alone (all P<0.05)
Abstract
Background Extrathyroidal extension (ETE) is an important prognostic factor in papillary thyroid carcinoma (PTC) that directly influences surgical decision-making, yet its accurate preoperative assessment by conventional ultrasound remains challenging. This study aimed to establish a nomogram integrating dual-plane (transverse and longitudinal combined) ultrasound radiomics and clinical features for the preoperative prediction of ETE in PTC. Methods A retrospective study was conducted on 483 patients with surgically confirmed PTC at our institution between January 2020 and December 2023 as the training cohort, and an additional 103 patients treated between January 2024 and June 2024 as the independent validation cohort. Radiomics features were extracted from transverse and longitudinal plane ultrasound images. The least absolute shrinkage and selection operator (LASSO) with five-fold cross-validation was applied for feature selection. Independent clinical predictors were identified by univariate and multivariate logistic regression analyses. A model integrating radiomics and clinical predictors for predicting ETE in PTC was constructed and presented as a nomogram. Results The combined model exhibited superior predictive performance, with areas under the receiver operating characteristic (ROC) curve (AUCs) of 0.880 in the training cohort and 0.858 in the independent validation cohort, and significantly outperformed both the clinical model and radiomics model alone (all P<0.05). The dual-plane radiomics strategy provided incremental value compared with single-plane models. Maximum tumor diameter and nodule-capsule contact were identified as independent clinical predictors (both P<0.05). Calibration curves and decision curve analysis (DCA) further verified the favorable consistency and clinical utility of the combined nomogram. Conclusions The nomogram integrating dual-plane ultrasound radiomics and clinical features provides a reliable, noninvasive tool for preoperative prediction of ETE in PTC, assisting in individualized risk stratification and clinical decision-making.