S. Mascharak1, A. L. Moore1,2, B. Duoto1, D. S. Foster1, R. E. Jones1,3, G. Wernig4, M. T. Longaker1 1Stanford University,Dept. Surgery,Palo Alto, CA, USA 2Brigham And Women’s Hospital,Dept. Surgery,Boston, MA, USA 3University Of Texas Southwestern Medical Center,Dept. Surgery,Dallas, TX, USA 4Stanford University,Dept. Pathology,Palo Alto, CA, USA
Introduction: Studies on dermal wound healing and scarring commonly rely on qualitative assessments of morphological and histological characteristics. These can be time consuming and prone to individual observer bias. Furthermore, it is not always feasible to visually ascertain metrics of scar organization on the micron scale. We have developed a package of image processing methods to automate quantification of morphological and histological dermal scar features. Using this tool, we elucidate the effects of an anti-scarring treatment on collagen content, fiber organization, and hair follicle neogenesis after dermal wounding.
Methods: Two stented dermal wounds were made on the dorsa of C57BL/6 mice and injected with various concentrations (PBS, 0.1, 2, 20 mg/mL) of a potential anti-scarring drug (Doxycycline). Histology (Masson’s Trichrome, Picrosirius Red) and hair follicle neogenesis were assessed at 14 and 28 days, respectively, and analyzed with Matlab 2017a. Histological features were measured using a combination of color deconvolution, adaptive filtering, and skeletonization of individual collagen fibers. Hair follicles were segmented using top-hat and shape filters.
Results: Automated quantification of color-deconvoluted histology images showed a significant decrease in collagen deposition for wounds treated with anti-scarring drug (*p = 0.036, 2 mg/mL vs PBS, Fig. 1A-C). Color deconvolution also segmented mature (red) and immature (green) collagen fibers in birefringent Picrosirius Red-stained samples, revealing a concomitant increase in immature collagen deposition (*p = 0.018, 2 mg/mL vs PBS, Fig. 1D Picro Green). Next, images of collagen fibers were skeletonized to trace individual tracks, revealing significant decreases in mature fiber length (*p = 0.042, 2 mg/mL vs PBS), number (*p = 0.036), and branching (*p = 0.038) after treatment (Fig. 1E-F). Strikingly, alignment of mature collagen fibers decreased after treatment (*p = 0.048, 2 mg/mL vs PBS), indicating basket-weave organization reminiscent of unwounded skin. Shape filtration segmented hair follicles, dramatically speeding up the counting process (Fig. 1G). A significant decrease in follicles was observed after treatment (*p = 0.025, 2 mg/mL vs PBS), though no change in morphology was observed (Fig. 1H).
Conclusion: Automated image processing facilitates histological and morphological study of scars. Additionally, analysis of collagen fibers reveals differences in scar organization after treatment with Doxycycline, that would not be apparent to the naked eye. Based on these findings, we will expand the use of image processing tools to models of hypertrophic scarring and fetal wound healing, towards the goal of new quantitative benchmarks for scar analysis.
