Development of an Automatic Algorithm Enabling Layer Segmentation and Optical Characteristic Analysis in Skin Optical Coherence Tomography Imaging

Several optical technologies have been demonstrated as potential diagnostic tools for skin imaging, such as reflectance confocal microscopy (RCM) and optical coherence tomography(OCT). Although RCM could support cellular-level imaging, OCT could provide a larger field of view and a longer imaging depth than RCM. To quantitatively evaluate the skin condition with OCT, it is essential to develop an algorithm extracting different features, for example, the epidermis thickness and the optical characteristic of the epidermis. Although various segmentation algorithms have been proposed, most of the golden standards used involve the delineation of the boundary manually, where the labeling is highly relied on clinicians' experience and might vary among different physicians and the physician him or herself. Therefore, in this study, we collected skin OCT images of different sites from 20 subjects using a portable spectral-domain OCT system. The contrast of the OCT images can be effectively improved by optical attenuation coefficient (OAC) computation. This enables the development of a fully automatic segmentation algorithm, providing the parameters such as the (i) epidermis thickness and the (ii, iii) roughness of the boundary between the epidermis and air as well as the epidermis and dermis. Also, the (iv, v) OAC coefficients of the epidermis and upper dermis layer are available. Collectively, the developed algorithm supports the quantitative analysis of the five parameters across the imaging site with volumetric OCT imaging. We believe the developed algorithm can facilitate the implementation of skin OCT imaging for aesthetic medicine as a modality for objective pre- and post-treatment evaluation.