Related Experiment Video
Updated: Aug 13, 2026
09:31
In Vivo, Percutaneous, Needle Based, Optical Coherence Tomography of Renal Masses
Published on: March 30, 2015
A Three-Slice Deep Learning-Radiomics Nomogram for Challenging Renal Mass Cases: A Double-Center Study
Yu Shu1,2, Hangzhe Sun3, Jiayue Zhang3,4
1Department of Ultrasonic Diagnosis, The Second Affiliated Hospital and Yuying Children's Hospital of Wenzhou Medical University, Wenzhou, Zhejiang, 325027, China.
Journal of Imaging Informatics in Medicine
|August 11, 2026
Summary
A 2.5D deep learning-radiomics approach effectively differentiates fat-poor angiomyolipoma from clear cell renal cell carcinoma. An integrated nomogram shows robust performance in challenging cases, aiding diagnostic uncertainty resolution.
Area of Science:
- Radiology
- Artificial Intelligence
- Oncology
Background:
- Differentiating fat-poor angiomyolipoma (fp-AML) from clear cell renal cell carcinoma (ccRCC) is challenging.
- Accurate differentiation is crucial for appropriate patient management.
Purpose of the Study:
- To identify the optimal spatial-depth architecture for deep learning-radiomics (DLR) in fp-AML versus ccRCC differentiation.
- To validate an integrated nomogram using diagnostically equivocal or discordant cases.
Main Methods:
- Retrospective study of 583 patients across two centers.
- Comparison of 2D (single-slice) and 2.5D (three-slice) DLR pipelines using a ResNet-50 backbone.
- Integration of the optimal DLR signature with clinical predictors into a nomogram, validated on challenging cases.
Main Results:
- The 2.5D DLR model outperformed the 2D model in both internal (AUC 0.821 vs. 0.782) and external (AUC 0.809 vs. 0.763) stress test cohorts.
- The integrated nomogram achieved high discriminative performance (internal AUC 0.855; external AUC 0.823).
- Decision curve and calibration analyses confirmed clinical utility and reliability.
Conclusions:
- The 2.5D architecture is optimal for DLR-based fp-AML vs. ccRCC differentiation.
- The integrated nomogram serves as a valuable decision support tool for challenging diagnostic scenarios.