JoVE - Journal of Visualized Experiments
JoVE Visualize
Contact Us

Model Based on Ultrasound Radiomics and Machine Learning to Preoperative Differentiation of Follicular Thyroid Neoplasm

  • 0Department of Ultrasound, The Second Affiliated Hospital, Jiangxi Medical College, Nanchang University, Nanchang, China.
Journal of Ultrasound in Medicine : Official Journal of the American Institute of Ultrasound in Medicine +

|

November 18, 2024

|

November 18, 2024

View abstract on PubMed

Summary

This summary is machine-generated.

This study shows that ultrasound radiomics can accurately differentiate follicular thyroid carcinoma (FTC) from follicular thyroid adenoma (FTA). A combined model offers a noninvasive tool for preoperative identification of these thyroid conditions.

Area Of Science

  • Medical Imaging
  • Oncology
  • Artificial Intelligence

Background

  • Distinguishing follicular thyroid carcinoma (FTC) from follicular thyroid adenoma (FTA) is crucial for patient management.
  • Accurate preoperative differentiation can prevent unnecessary surgeries for benign conditions.

Purpose Of The Study

  • To evaluate the diagnostic value of ultrasound radiomics in differentiating FTC from FTA.
  • To develop a noninvasive preoperative prediction tool for FTC and FTA.

Main Methods

  • Retrospective analysis of ultrasound images and clinical data from 389 patients across three institutions.
  • Development of radiomics models using machine learning classifiers, including a combined model with clinical characteristics.
  • Performance evaluation using receiver operating characteristic curves, calibration, and decision curves.

Main Results

  • The random forest-based radiomics model demonstrated strong performance in differentiating FTC and FTA (AUCs ranging from 0.821 to 0.880 across cohorts).
  • The combined model, integrating radiomics and clinical features, showed superior efficacy (AUCs ranging from 0.874 to 0.883).
  • Good consistency and high clinical benefit were observed for the combined model.

Conclusions

  • Ultrasound radiomics, particularly with a random forest approach, is a feasible method for differentiating FTC and FTA.
  • The developed combined model serves as an effective, noninvasive tool for preoperative identification of FTC and FTA.

Keywords: