The ability to image the physiology of microvasculature with high spatial resolution in three dimensions while investigating structural changes of skin, is essential for understanding the complex processes of skin aging, wound healing and disease development. Further, the quantitative, automatic assessment of these changes enables to analyze large amounts of image data in an abstract but comprehensive manner.
However, previous work using OCT with methods of angiography was imaging less scattering, hence more challenging tissue than skin, such as brain and retina tissue. The published methods for capillary segmentation were mostly non-automatic, poorly benchmarked against state-of-the-art methods of computer vision and not applied to investigate medical processes and studies in a comprehensive fashion.
Here, segmentation of capillaries in skin is reported and its efficacy is demonstrated in both, a
longitudinal mouse study and a preliminary study in humans. By combining state-of-the-art image
processing methods in an optimized way, we were able to improve the segmentation results and analyze the impact of each post-processing step.
Furthermore, this automatic segmentation enabled us to analyze big amounts of
datasets automatically and derive meaningful conclusions for the planning of clinical studies.
With this work, optical coherence tomography is combined with methods of computer vision to a diagnostic
tool with unique capabilities to characterize vascular diversity and provide extraordinary
opportunities for dermatological investigation in both, clinics and research.