Related Experiment Video
Updated: Sep 6, 2025
07:15
Machine Learning Algorithms for Early Detection of Bone Metastases in an Experimental Rat Model
Published on: August 16, 2020
6.9K
Machine learning-based radiomics for histological classification of parotid tumors using morphological MRI: a
Zhiying He1,2,3, Yitao Mao4, Shanhong Lu1,2,3
1Department of Otolaryngology Head and Neck Surgery, Xiangya Hospital, Central South University, 87 Xiangya Road, Changsha, Hunan, 410008, People's Republic of China.
European Radiology
|June 24, 2022
Summary
Machine learning models using magnetic resonance imaging (MRI) radiomics can help classify parotid tumors. These AI tools show potential in assisting radiologists with preliminary diagnosis.
Area of Science:
- Radiomics and Artificial Intelligence in Medical Imaging
- Oncology and Diagnostic Radiology
Background:
- Parotid tumors require accurate classification for effective treatment.
- Morphological magnetic resonance imaging (MRI) provides rich data for analysis.
- Machine learning (ML) offers advanced analytical capabilities for medical data.
Purpose of the Study:
- To assess the efficacy of ML models utilizing morphological MRI radiomics for parotid tumor classification.
- To compare the diagnostic performance of ML models against human radiologists.
Main Methods:
- 298 parotid tumor patients were divided into training (70%) and testing (30%) sets.
- Radiomics features were extracted from morphological MRI and selected using Select K Best and LASSO algorithms.
- XGBoost, Support Vector Machine (SVM), and Decision Tree (DT) models were developed for four-subtype classification, with performance evaluated by ROC curves and confusion matrices.
Main Results:
- Optimal feature sets of 6, 12, and 8 were identified in a three-step process.
- XGBoost demonstrated the highest Area Under the Curve (AUC) in training and early testing phases.
- XGBoost and SVM models achieved higher accuracies (70.8% and 59.6%) than DT (46.1%) and radiologists (49.2%) in the test cohort.
Conclusions:
- ML models leveraging morphological MRI radiomics show promise as assistive tools for parotid tumor classification.
- XGBoost and SVM models outperformed radiologists in classifying parotid tumor subtypes using only morphological MRI.
- These AI-driven approaches can serve as valuable adjuncts in clinical practice, particularly for preliminary screening.
