Systemic sclerosis & Raynaud's
Artificial Intelligence–Aided Lung Ultrasound Detection of Interstitial Lung Disease in Systemic Sclerosis and Inflammatory Myopathy
Authors' conclusion
Deep learning applied to LUS enables accurate ILD detection in connective tissue disease and can enhance expert interpretation.
Study context
- Population
- Patients with systemic sclerosis (SSc) or idiopathic inflammatory myopathy (IIM) ± interstitial lung disease (ILD) with paired lung ultrasound (LUS) and chest computed tomography (CT)
- Participants
- 140 patients3,920 LUS images
- Comparison
- CNN models (InceptionV3, ResNet-50, VGG-16, LUS-Net) vs CT and LUS-ILD-24 human interpretation for ILD detection on LUS
- Outcome
- ILD detection performance at image and patient levels (AUC, sensitivity, specificity, agreement with expert interpretation)
Methodological detailsdeep learning (CNN, transfer learning), development and independent test sets, paired LUS and chest CT
- Deep learning (CNN, transfer learning)
- Development and independent test sets
- Paired LUS and chest CT
Performance
As reported in the abstract
Patient-level ILD detection
| Test or model | AUC | Sensitivity | Specificity |
|---|---|---|---|
| VGG-16 | 0.972 | 97.4% | 92.6% |
VGG-16
AUC0.972
Sensitivity97.4%
Specificity92.6%
Further findings
Grad-CAM regions driving predictions
- Pleural features were the primary regions influencing model predictions
Original publication
Artificial Intelligence–Aided Lung Ultrasound Detection of Interstitial Lung Disease in Systemic Sclerosis and Inflammatory Myopathy. Arthritis Care Res (Hoboken). Published 6 April 2026. doi:10.
Composed by AI from the published abstract and checked against it. How Briefs are made →