The value of radiomic models based on contrast-enhanced computed tomography images in distinguishing between pancreatic ductal adenocarcinoma and pancreatic neuroendocrine neoplasms
Objective To evaluate the value of radiomic models based on contrast-enhanced computed tomography (CT) images in differentiating pancreatic ductal adenocarcinoma (PDAC) from pancreatic neuroendocrine neoplasms (pNEN). Methods A total of 291 cases of pancreatic tumors (218 PDAC cases and 73 pNEN case...
Saved in:
| Main Author: | |
|---|---|
| Format: | Article |
| Language: | zho |
| Published: |
Editorial Office of Journal of Precision Medicine
2025-06-01
|
| Series: | 精准医学杂志 |
| Subjects: | |
| Online Access: | https://jpmed.qdu.edu.cn/fileup/2096-529X/PDF/1750385771262-1699879288.pdf |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Objective To evaluate the value of radiomic models based on contrast-enhanced computed tomography (CT) images in differentiating pancreatic ductal adenocarcinoma (PDAC) from pancreatic neuroendocrine neoplasms (pNEN). Methods A total of 291 cases of pancreatic tumors (218 PDAC cases and 73 pNEN cases) from The Affiliated Hospital of Qingdao University from November 2016 to September 2022 were assigned to the training set, while 124 cases (93 PDAC cases and 31 pNEN cases) from Shandong Provincial Hospital were assigned to the test set. Tumor regions of interest (ROI) were delineated on arterial-phase and venous-phase contrast-enhanced CT images of the upper abdomen, and subsequently, radiomic features were extracted and selected. Radom forest models were constructed using the radiomic features preprocessed with six methods separately: L1-norm regularization, L2-norm regularization, maximum absolute normalization, min-max normalization, quantile transformation, and Yeo-Johnson transformation. The efficacy of the six random forest models for distinguishing between PDAC and pNEN was assessed using the area under the receiver operating characteristic curve (AUC). The Brier score (BS) was calculated to evaluate model calibration. A decision curve analysis (DCA) was performed to quantify the net clinical benefit. Results For the six random forest models, the AUCs were 0.981-0.998 in the training set and 0.685-0.840 in the test set, with the accuracies being 92.8%-98.2% and 51.8%-86.7%, respectively. The quantile transformation-based random forest model demonstrated the best perfor-mance, showing an AUC of 0.840, a sensitivity of 90.2%, a specificity of 77.0%, and an accuracy of 86.7%, significantly surpassing those of the other models, also with the best calibration (BS=0.119). The DCA results indicated that the quantile transformation-based random forest model demonstrated superior net clinical benefits in both the training set and test set. Conclusion Radiomic models based on contrast-enhanced CT features can effectively diffe-rentiate PDAC from pNEN, in which the quantile transformation-based random forest model exhibits superior diagnostic performance, demonstrating potential for clinical application. |
|---|---|
| ISSN: | 2096-529X |