Related Experiment Video
Updated: Jun 28, 2025
03:55
Computer-Aided Three-Dimensional Visualization in the Treatment of Locally Advanced Thyroid Cancer
Published on: June 9, 2023
524
Cervical lymph node metastasis prediction from papillary thyroid carcinoma US videos: a prospective multicenter study
Ming-Bo Zhang1, Zhe-Ling Meng2, Yi Mao1
1Department of Ultrasound, the First Medical Center, General Hospital of Chinese PLA, Beijing, China.
BMC Medicine
|April 12, 2024
Summary
Artificial intelligence (AI) accurately predicts lymph node metastasis in papillary thyroid carcinoma using ultrasound videos. This AI tool enhances radiologist performance, improving diagnostic accuracy and sensitivity for better patient management.
Area of Science:
- Oncology
- Medical Imaging
- Artificial Intelligence
Background:
- Accurate prediction of lymph node metastasis (LNM) is crucial for tailoring treatment in papillary thyroid carcinoma (PTC) patients.
- Current methods risk overtreatment or undertreatment; artificial intelligence (AI) trained on thyroid ultrasound (US) shows promise for improved prediction.
Purpose of the Study:
- To develop and validate a deep learning model for predicting LNM in PTC using US videos.
- To assess the AI model's diagnostic performance and its auxiliary effect on radiologists.
Main Methods:
- A multi-scale, multi-frame, dual-direction deep learning (MMD-DL) model was pre-trained on retrospective PTC patient data (n=488).
- The MMD-DL model was fine-tuned and prospectively validated across four centers using additional PTC patient data (n=218).
- Diagnostic performance was evaluated using ROC curves, AUC, accuracy, sensitivity, and specificity, comparing AI-assisted radiologists to unassisted ones.
Main Results:
- The MMD-DL model achieved AUCs of 0.85 in the test cohort and 0.81 in the validation cohort.
- Radiologists using the AI model showed significant improvements in average diagnostic accuracy (60% vs. 57%) and sensitivity (65% vs. 62%).
Conclusions:
- AI models trained on US videos can accurately and reproducibly predict cervical LNM in PTC patients preoperatively.
- The AI model serves as an effective tool to enhance the diagnostic performance of US radiologists.
