Systemic sclerosis & Raynaud's
External clinical validation of automated software to identify structural abnormalities and microhaemorrhages in nailfold videocapillaroscopy images
Authors' conclusion
This external clinical validation suggests that this algorithm is useful to assist in the diagnosis and follow-up of SSc or RP patients in a timely manner.
Study context
- Population
- Nailfold videocapillaroscopy (NVC) images of Raynaud's phenomenon (RP) patients
- Participants
- 1,164 NVC images
- Comparison
- Automated algorithm vs consensus annotation of trained capillaroscopists
- Outcome
- Matches and discrepancies between algorithm predictions and annotations by consensus of ≥3 or ≥4 interobservers
Methodological detailsexternal clinical validation, deep convolutional neural network-based algorithm
- External clinical validation
- Deep convolutional neural network-based algorithm
Performance
As reported in the abstract
Agreement of algorithm with expert consensus
| Test or model | Correctly predicted, images with consensus among ≥3 capillaroscopists (consensus in 86.9% of images) | Matched expert panel, cases with consensus among ≥4 experts (consensus in 52.0% of cases) | PPV for microhaemorrhages and unaltered, giant or abnormal capillaries | Sensitivity for dilations and tortuosities | NPV and specificity, all categories |
|---|---|---|---|---|---|
| Algorithm | 75.8% | 87.1% | >80% | >75% | >89% |
Algorithm
Correctly predicted, images with consensus among ≥3 capillaroscopists (consensus in 86.9% of images)75.8%
Matched expert panel, cases with consensus among ≥4 experts (consensus in 52.0% of cases)87.1%
PPV for microhaemorrhages and unaltered, giant or abnormal capillaries>80%
Sensitivity for dilations and tortuosities>75%
NPV and specificity, all categories>89%
Original publication
External clinical validation of automated software to identify structural abnormalities and microhaemorrhages in nailfold videocapillaroscopy images. Clin Exp Rheumatol. Published 1 December 2022. doi:10.
Composed by AI from the published abstract and checked against it. How Briefs are made →