Related Experiment Video
Updated: Aug 5, 2026
10:02
High-frequency Ultrasound Imaging of Mouse Cervical Lymph Nodes
Published on: July 25, 2015
A Multimodal Ultrasound Radiomics Model Integrating Contrast-Enhanced Phases: Retrospective Temporal Test Cohort
Mingru Gao1, Qian Lv1, Haohao Hao1
1Department of Ultrasound, The Fourth Affiliated Hospital of Harbin Medical University, Harbin 150001, Heilongjiang, China (M.G., Q.L., H.H., N.Z., Y.G., K.M.).
Academic Radiology
|August 3, 2026
Summary
A novel multimodal ultrasound radiomics model integrating contrast-enhanced ultrasound (CEUS) shows promising results for distinguishing benign from malignant cervical lymph nodes. This AI-driven approach may aid clinicians in lymph node assessment.
Area of Science:
- Radiology
- Artificial Intelligence
- Oncology
Background:
- Cervical lymph node assessment is crucial for cancer staging.
- Distinguishing benign from malignant lymph nodes can be challenging.
- Advanced imaging techniques can improve diagnostic accuracy.
Purpose of the Study:
- To develop a multimodal ultrasound radiomics model for differentiating benign from malignant cervical lymph nodes.
- To evaluate the model's performance against sonographers with varying experience levels.
Main Methods:
- A retrospective study involving 245 patients in the development cohort and 60 in a temporal test cohort.
- Extraction of deep features from conventional ultrasound and three-phase contrast-enhanced ultrasound (CEUS) key frames using ResNet50.
- Development and validation of a support vector machine (SVM) model integrated with CEUS data.
Main Results:
- The CEUS-integrated SVM model achieved an AUC of 0.851 in the temporal test cohort.
- The model demonstrated higher sensitivity than senior and junior sonographers.
- Model performance was comparable to experienced sonographers, with potential as an adjunct tool.
Conclusions:
- A multimodal ultrasound radiomics model incorporating CEUS exhibits feasible diagnostic performance for cervical lymph node assessment.
- The model shows potential as a hypothesis-generating adjunct tool.
- Further multi-center evaluation is needed before clinical application due to the retrospective, single-center design.
