The following is a summary of “A deep learning system for quantitative assessment of microvascular abnormalities in nailfold capillary images,” published in the June 2023 issue of Rheumatology by Bharathi, et al.
For a study, researchers sought to develop and validate a fully automated image analysis system for nailfold capillaroscopy in rheumatology clinics to aid in the timely diagnosis of systemic sclerosis (SSc).
Researchers utilized deep learning networks to detect each capillary in the distal row of vessels and perform morphological measurements, following the image interpretation strategies of SSc experts. The system provided a subject-level probability of SSc by combining measurements from multiple fingers. The system was evaluated using photographs from various individuals, including 66 images taken using a low-cost digital microscope (group C) and 132 images taken at high resolution (group B). The system was trained using high-quality images from 111 subjects (group A).
Approximately half of each group had confirmed SSc, while the other half were healthy controls or had primary Raynaud’s phenomenon (RP). The performance of SSc experts was also estimated for comparison.
The automated system’s diagnostic accuracy was evaluated by comparing the SSc probabilities with the known clinical status of patients (SSc versus ‘normal’), generating receiver operating characteristic curves (ROCs). For group B, the area under the ROC (AUC) was 97% (median, 90% CI: 94–99%), with equal sensitivity and specificity of 91% (86–95%). In group C, the AUC was 95% (88–99%), with equal sensitivity and specificity of 89% (82–95%). In comparison, SSc expert consensus achieved a sensitivity of 82% and specificity of 73%.
The fully automated image analysis system, employing deep learning, demonstrated diagnostic performance at least as good as SSc experts. Moreover, it proved robust and effective even with images captured using a low-cost digital microscope. The automated system has the potential to improve nailfold capillaroscopy utilization in rheumatology clinics for the early detection of SSc.
Source: academic.oup.com/rheumatology/article/62/6/2325/6991166