Related Experiment Video
Updated: Jul 20, 2025
04:23
A Swin Transformer-Based Model for Thyroid Nodule Detection in Ultrasound Images
Published on: April 21, 2023
1.9K
Predicting BRAFV600E mutations in papillary thyroid carcinoma using six machine learning algorithms based on
Enock Adjei Agyekum1,2, Yu-Guo Wang3, Fei-Ju Xu1
1Ultrasound Medical Laboratory, Department of Ultrasound, Affiliated People's Hospital of Jiangsu University, Zhenjiang, 212002, China.
Scientific Reports
|August 3, 2023
Summary
Machine learning models using ultrasound elastography radiomic features can predict the BRAF V600E mutation in papillary thyroid cancer (PTC) patients. The support vector machine with radial basis function kernel demonstrated the highest accuracy, aiding in pre-surgical risk assessment.
Area of Science:
- Radiology and Medical Imaging
- Oncology
- Biomedical Engineering
Background:
- The BRAF V600E mutation is a common driver in papillary thyroid cancer (PTC), leading to uncontrolled cell proliferation.
- Accurate pre-surgical identification of BRAF V600E mutation status is crucial for treatment planning.
Purpose of the Study:
- To develop and validate machine learning algorithms for predicting BRAF V600E mutation in PTC patients using ultrasound (US) elastography radiomic features.
- To compare the performance of six different machine learning models in this predictive task.
Main Methods:
- Retrospective analysis of US strain elastography images from 138 PTC patients.
- Extraction of 479 radiomic features, followed by feature selection using Pearson's Correlation Coefficient and Recursive Feature Elimination.
- Training and validation of six machine learning algorithms (SVM_L, SVM_RBF, LR, NB, KNN, LDA) to predict BRAF V600E mutation.
Main Results:
- The study identified 27 key radiomic features for predicting BRAF V600E mutation.
- The support vector machine with radial basis function kernel (SVM_RBF) achieved the highest performance with an accuracy of 0.93 and an AUC of 0.98.
- Significant differences in echogenicity, diameter ratios, and elasticity were observed between PTC patients with and without the BRAF V600E mutation.
Conclusions:
- Machine learning models utilizing US elastography radiomic features can effectively predict BRAF V600E mutation status in PTC patients.
- The SVM_RBF model shows high potential for assisting clinicians in pre-operative risk stratification for PTC.
- This approach offers a non-invasive method to potentially guide surgical and treatment decisions.
