MR radiomics for predicting lymph node metastasis of pancreatic ductal adenocarcinoma by comparing different contrast-enhanced phases
- 1Department of Radiology, Affiliated Hospital of North Sichuan Medical College, Nanchong, 637000 China
- 2Department of Radiology, Chongqing University Cancer Hospital & Chongqing Cancer Institute & Chongqing Cancer Hospital, Chongqing, P.R. China
Abstract
Background: Lymph node metastasis (LNM) is an important factor affecting the prognosis of pancreatic ductal adenocarcinoma (PDAC). However, there is a lack of effective methods to predict LNM. The purpose of this study was to investigate the potential value of MR radiomics to predict LNM of PDAC preoperatively by comparing the different phases of contrast-enhanced T1 weight imaging (T1WI).
Methods: A number of 123 patients with pathologically confirmed PDAC were involved in this study and divided into training and validation cohorts based on a 7:3 ratio. A radiologist with ten years of experience evaluated LNM based on morphological abnormalities of MR images. Radiomics scores (Rad-score) of arterial phase (AP), pancreatic phase (PP), and venous phase (VP) were calculated by multiplying the sum of each feature by a non-zero factor of the least absolute shrinkage and selection operator. We then established the nomogram integrating the VP Rad-score and the tumor location using logistic regression. Predictive ability was assessed using receiver operating characteristic (ROC) testing, area under the curve (AUC), accuracy, specificity, and sensitivity. The nomogram’s performance was evaluated using the calibration curve analysis, Hosmer-Lemeshow test and decision curve analysis (DCA). A comparison of AUC was performed among the Rad-score of the three phases using the DeLong test.
Results: The AUC of the radiologist’s evaluation was 0.590 in the training cohort and 0.411 in the validation cohort. The AUC of AP Rad-score, PP Rad-score, and VP Rad-score was 0.778, 0.776, and 0.946 in the training cohort and 0.753, 0.735, and 0.961 in the validation cohort, respectively. DeLong test demonstrated that the Rad-score of three phases had higher AUC than the radiologist’s evaluation (AP Rad-score VS radiologist’s evaluation [p-value = 0.005], PP Rad-score VS radiologist’s evaluation [p-value = 0.003], VP Rad-score VS radiologist’s evaluation [p-value = 0.001]). Besides, the AUC of the VP Rad-score was higher than that of the AP Rad-score and the PP Rad-score (AP Rad-score VS VP Rad-score [p-value = 0.001], PP Rad-score VS VP Rad-score [p-value = 0.001]). Both the calibration test and DCA showed the nomogram was effective and reliable, suggesting its potential clinical utility.
Conclusions: The Radiomics of VP of contrast-enhanced T1WI may be a promising method for predicting LNM of PDAC, which can provide a reference for the selection of treatment plans and prognosis evaluation of PDAC patients.
Related
- Habitat radiomics combined with clinical features for predicting tertiary lymphoid structure status in pancreatic ductal adenocarcinoma: development and validation of a preoperative CECT-based model.
- The clinical value of ultrasound radiomics combined with clinical data nomogram in predicting cervical lymph node metastasis of papillary thyroid carcinoma.
- CT imaging-based radiomics and deep learning models for predicting chemotherapy response in advanced pancreatic cancer
- An explainable machine learning model combining intratumoral and peritumoral computed tomography radiomics with maximum standard uptake value for predicting International Association for the Study of Lung Cancer grading and prognosis in stage I pure-solid lung adenocarcinoma.
- A Multidimensional CT-Based Tumor Location Metric for Predicting Lymph Node Metastasis and Occult Lymph Node Metastasis in Lung Adenocarcinoma
BibTeX
@article{liu2026radiomics,
title = {{MR radiomics for predicting lymph node metastasis of pancreatic ductal adenocarcinoma by comparing different contrast-enhanced phases}},
author = {Zhuoyu Liu and Nian Liu and Yuwei Chen and Qing Jia and Qinglin Du and Xiaohua Huang},
journal = {BMC Medical Imaging},
year = {2026},
month = aug,
doi = {10.1186/s12880-026-02665-w},
url = {https://doi.org/10.1186/s12880-026-02665-w},
note = {CC BY-NC-ND 4.0},
}
RIS
TY - JOUR
TI - MR radiomics for predicting lymph node metastasis of pancreatic ductal adenocarcinoma by comparing different contrast-enhanced phases
AU - Zhuoyu Liu
AU - Nian Liu
AU - Yuwei Chen
AU - Qing Jia
AU - Qinglin Du
AU - Xiaohua Huang
PY - 2026
DA - 2026/08/07
JO - BMC Medical Imaging
DO - 10.1186/s12880-026-02665-w
UR - https://doi.org/10.1186/s12880-026-02665-w
AB - Background: Lymph node metastasis (LNM) is an important factor affecting the prognosis of pancreatic ductal adenocarcinoma (PDAC). However, there is a lack of effective methods to predict LNM. The purpose of this study was to investigate the potential value of MR radiomics to predict LNM of PDAC preoperatively by comparing the different phases of contrast-enhanced T1 weight imaging (T1WI). Methods: A number of 123 patients with pathologically confirmed PDAC were involved in this study and divided into training and validation cohorts based on a 7:3 ratio. A radiologist with ten years of experience evaluated LNM based on morphological abnormalities of MR images. Radiomics scores (Rad-score) of arterial phase (AP), pancreatic phase (PP), and venous phase (VP) were calculated by multiplying the sum of each feature by a non-zero factor of the least absolute shrinkage and selection operator. We then established the nomogram integrating the VP Rad-score and the tumor location using logistic regression. Predictive ability was assessed using receiver operating characteristic (ROC) testing, area under the curve (AUC), accuracy, specificity, and sensitivity. The nomogram’s performance was evaluated using the calibration curve analysis, Hosmer-Lemeshow test and decision curve analysis (DCA). A comparison of AUC was performed among the Rad-score of the three phases using the DeLong test. Results: The AUC of the radiologist’s evaluation was 0.590 in the training cohort and 0.411 in the validation cohort. The AUC of AP Rad-score, PP Rad-score, and VP Rad-score was 0.778, 0.776, and 0.946 in the training cohort and 0.753, 0.735, and 0.961 in the validation cohort, respectively. DeLong test demonstrated that the Rad-score of three phases had higher AUC than the radiologist’s evaluation (AP Rad-score VS radiologist’s evaluation [p-value = 0.005], PP Rad-score VS radiologist’s evaluation [p-value = 0.003], VP Rad-score VS radiologist’s evaluation [p-value = 0.001]). Besides, the AUC of the VP Rad-score was higher than that of the AP Rad-score and the PP Rad-score (AP Rad-score VS VP Rad-score [p-value = 0.001], PP Rad-score VS VP Rad-score [p-value = 0.001]). Both the calibration test and DCA showed the nomogram was effective and reliable, suggesting its potential clinical utility. Conclusions: The Radiomics of VP of contrast-enhanced T1WI may be a promising method for predicting LNM of PDAC, which can provide a reference for the selection of treatment plans and prognosis evaluation of PDAC patients.
SN - 1471-2342
ER -