Enhancing Tenosynovitis Diagnosis With XGBoost: A Machine Learning Approach Using Ultrasound Data.

Area of Science:

• Medical Imaging
• Machine Learning
• Rheumatology

Background:

• Rheumatoid arthritis (RA) is a systemic autoimmune disease causing joint destruction.
• Tenosynovitis is an early indicator of RA, necessitating accurate diagnostic tools.
• Objective diagnosis of tenosynovitis is crucial for timely RA evaluation and management.

Purpose of the Study:

• To develop and validate a machine learning (ML) model for objective tenosynovitis diagnosis in RA patients using ultrasound (US) radiomics.
• To assess the model's performance against conventional ML methods like Support Vector Machine (SVM) and Random Forest (RF).
• To evaluate the clinical utility and real-time applicability of the developed ML model.

Main Methods:

• Extraction of radiomic features from 1496 grayscale US images of tendons from 152 RA patients.
• Training an XGBoost ML model using radiomic and clinical features, with an 80/20% patient-level train-test split.
• Comparative analysis with SVM and RF, evaluating metrics including AUC, sensitivity, specificity, and processing time.

Main Results:

• The XGBoost model achieved high performance with an AUC of 0.914 on the test set.
• Demonstrated superior performance compared to SVM and RF models on both training and test datasets.
• Achieved a mean processing time of 0.8 seconds per case, indicating real-time clinical applicability.

Conclusions:

• The XGBoost-based ML model effectively diagnoses tenosynovitis in RA patients via US imaging.
• The model's superior performance and real-time capability suggest its potential for accurate RA evaluation.
• This approach offers an objective and efficient tool for managing rheumatoid arthritis.