Skip to main content
Fig. 2 | Arthritis Research & Therapy

Fig. 2

From: High-throughput quantitative histology in systemic sclerosis skin disease using computer vision

Fig. 2

Association between DNN-derived signatures and mRSS. A) To visualize the distribution of Quantitative Image Features (QIFs) for the primary cohort, a heat map was generated with columns consisting of image patches (26 patients × 100 image patches) and rows consisting of 4096 QIFs. For display, the rows and columns of the heat map were sorted by the first principal component, which highlights both the correlation among features across the set of image patches and the differences in QIF signatures across image patches. B) The distribution of the 100 Image Patch Scores for two biopsies is shown for a patient with low (orange), and high (blue) modified Rodnan Skin Score (mRSS). The mean Image Patch Score for each biopsy, termed the Biopsy Score (dotted vertical line) is shown. C) Biopsy Scores significantly correlated with mRSS (R = 0.55, p = 0.010, repeated measures test)

Back to article page
\