Related Experiment Video
Updated: Dec 28, 2025
05:41
Author Spotlight: Integrating Ultrasound Imaging with Biochemical Markers for Thyroid Disease Diagnosis
Published on: February 9, 2024
974
Multi-Reader Multi-Case Study for Performance Evaluation of High-Risk Thyroid Ultrasound with Computer-Aided
Ming-Hsun Wu1, Kuen-Yuan Chen1, Shyang-Rong Shih2
1Department of Surgery, National Taiwan University Hospital, Taipei 10002, Taiwan.
Cancers
|February 12, 2020
Summary
Computer-aided detection (CAD) systems improve physician accuracy in diagnosing thyroid cancer. This technology enhances diagnostic interpretation and reduces variability, aiding in better patient evaluation for potential malignancy.
Area of Science:
- Medical Imaging
- Artificial Intelligence in Medicine
- Oncology
Background:
- Thyroid cancer diagnosis relies on sonographic characteristics.
- Computer-aided detection (CAD) systems offer potential improvements in diagnostic accuracy.
Purpose of the Study:
- To evaluate the effectiveness of a CAD system in assisting physicians with tentative thyroid cancer diagnoses.
- To compare clinician performance with and without CAD assistance.
Main Methods:
- A multicenter, multireader, and multicase (MRMC) study involving 19 physicians.
- Physicians assessed 265 thyroid nodules from 200 patients, with and without CAD assistance.
- Analysis of interobserver variability and diagnostic performance using area under the curve (AUC).
Main Results:
- The average AUC significantly increased from 0.728 without CAD to 0.792 with CAD.
- CAD use decreased the standard deviation of malignancy scores (18.97 to 16.29) and the mean score for benign cases (35.01 to 31.24).
- CAD identified additional malignant nodules for evaluation (7.6%) and reduced unnecessary biopsies for benign nodules (10.8%).
Conclusions:
- CAD systems enhance clinicians' interpretation of thyroid nodules, improving diagnostic accuracy.
- The application of CAD systems can significantly lessen variability in thyroid cancer diagnosis.
- Further research is needed to assess CAD system utility in real-world ultrasound diagnostics with a higher prevalence of benign nodules.
