Epidermal layer recognition of optical coherence tomography skin images

Abstract

Epidermal layer recognition is an effective way to diagnose the severity of psoriasis and other skin diseases. Optical coherence tomography (OCT) is a non-invasive imaging modality that acquires images of tissue in vivo and plays an important role in detecting skin diseases and assessing therapeutic effects. In this paper, a method for epidermal layer recognition of OCT skin images is proposed. Firstly, the skin surface is obtained based on the axial gradient map. Then the original image is flattened according to the estimated skin surface. And the potential region of the dermo-epidermal junction (DEJ) is obtained according to the mean column signal. Finally, the epidermal layer is identified by topology transmission and seed filling. The method can automatically segment the curved mastoid structure of the DEJ which is suitable for OCT images of skin with weakened signal intensity of dermis due to dermatitis or other diseases.