Development and clinical validation of an artificial intelligence based model for thyroid nodule malignancy risk assessment using C-TIRADS guidelines - Report - MDSpire
Advertisement
Development and clinical validation of an artificial intelligence based model for thyroid nodule malignancy risk assessment using C-TIRADS guidelines
Clinical Report: AI-driven Model for Assessing Malignancy Risk in Thyroid Nodules
Overview
An AI model based on C-TIRADS criteria demonstrated an overall accuracy of 86.2% in assessing malignancy risk in thyroid nodules.
Background
Thyroid nodules are prevalent, with a significant portion being malignant. Accurate assessment of these nodules is crucial for effective management, as benign nodules typically require observation while malignant ones necessitate surgical intervention. Traditional ultrasound evaluations can be subjective and variable, highlighting the need for standardized, objective assessment methods.
Data Highlights
Parameter
AI Model
Physician (Without AI)
Physician (With AI)
Accuracy
0.862 (95% CI, 0.822–0.901)
0.705 (95% CI, 0.657–0.756)
0.845 (95% CI, 0.802–0.884)
Key Findings
The AI model achieved an overall accuracy of 86.2% in malignancy risk assessment.
Physician accuracy improved from 70.5% to 84.5% with AI assistance.
The model is based on C-TIRADS ultrasound features including composition, echogenicity, margin, shape, and echogenic foci.
Standardized malignancy risk stratification can aid in thyroid ultrasound interpretation.
Larger multicenter studies are needed for routine clinical application.
Clinical Implications
The AI-driven model provides a tool for thyroid nodule assessments.
Conclusion
The C-TIRADS-guided AI framework shows potential in standardizing malignancy risk assessment for thyroid nodules.
In the first installment of a new series on the investment environment in cell and gene therapy, Circio CEO Erik Wiklund explains how he had to get creative to raise funds in a difficult biotech market
