Related Experiment Video
Updated: Jun 25, 2025
06:31
Author Spotlight: Enhancing Rheumatoid Arthritis Research Through HR-pQCT Imaging Analysis
Published on: October 6, 2023
2.1K
Fully automatic quantification for hand synovitis in rheumatoid arthritis using pixel-classification-based
Wanxuan Fang1, Yijun Mao1, Haolin Wang1
1Graduate School of Health Sciences, Hokkaido University, North-12 West-5, Kita-Ku, Sapporo, 060-0812, Japan.
Japanese Journal of Radiology
|May 24, 2024
Summary
This study introduces a deep learning method for analyzing dynamic contrast-enhanced MRI time-intensity curves to automatically quantify synovium in rheumatoid arthritis patients, showing high accuracy in segmenting enhanced pannus.
Area of Science:
- Medical Imaging
- Artificial Intelligence
- Rheumatology
Background:
- Rheumatoid arthritis (RA) diagnosis and monitoring often involve assessing synovitis, the inflammation of the synovial membrane.
- Dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) provides functional information about tissue perfusion and vascularity, useful for detecting active inflammation.
- Accurate quantification of synovium is crucial for evaluating disease activity and treatment response in RA patients.
Purpose of the Study:
- To develop and validate a deep learning (DL) based classification method for segmenting and quantifying enhanced synovium in the hands of RA patients using DCE-MRI.
- To leverage time-intensity curve (TIC) analysis from DCE-MRI data to differentiate between various tissue types and identify inflamed pannus.
Main Methods:
- A retrospective analysis of hand DCE-MRI scans from 28 RA patients was performed.
- A DL model incorporating dilated causal convolution and SELU activation was developed for enhanced pannus segmentation, trained on pixel data with corresponding TICs.
- Preprocessing included noise reduction, motion correction, contrast enhancement, and TIC normalization. Leave-one-out cross-validation was employed for model testing.
Main Results:
- The DL model achieved high performance in enhanced pannus segmentation, with a pixel-level sensitivity of 85%, specificity of 98%, accuracy of 99%, and precision of 84%, and a mean Dice score of 0.73.
- DL-measured enhanced pannus volume demonstrated strong correlations with ground truth at both pixel-based (r=0.87) and patient-based (r=0.84) levels (p < 0.001 for both).
- Bland-Altman analysis indicated good agreement between DL measurements and ground truth, with small mean differences in volume estimations.
Conclusions:
- The proposed DL-based DCE-MRI TIC analysis method shows significant potential for the automatic and accurate segmentation and quantification of enhanced synovium in RA patients.
- This approach could aid in objective disease assessment and monitoring in clinical practice.
