Optical Coherence Tomography (OCT) is a non-invasive imaging modality that acquires cross sectional images of tissue
in-vivo. It accelerates skin diagnosis by eliminating invasive biopsy and laborious histology in the process.
Dermatologists have widely used it for looking at morphology of skin diseases such as psoriasis, dermatitis, basal cell
carcinoma etc. Skin scientists have also successfully used it for looking at differences in epidermal thickness and its
underlying structure with respect to age, body sites, ethnicity, gender, and other related factors.
Similar to other in-vivo imaging systems, OCT images suffer from a high degree of speckle and noise content, which
hinders examination of tissue structures. Most of the previous work in OCT segmentation of skin was done manually.
This compromised the quality of the results by limiting the analyses to a few frames per area.
In this paper, we discuss a region growing method for automatic identification of the upper and lower boundaries of the
epidermis in living human skin tissue. This image analysis method utilizes images obtained from a frequency-domain
OCT. This system is high-resolution and high-speed, and thus capable of capturing volumetric images of the skin in
short time. The three-dimensional (3D) data provides additional information that is used in the segmentation process to
help compensate for the inherent noise in the images. This method not only provides a better estimation of the epidermal
thickness, but also generates a 3D surface map of the epidermal-dermal junction, from which underlying topography can
be visualized and further quantified.