Optik 124 (2013) 5665–5668 Contents lists available at ScienceDirect Optik jou rn al homepage: www.elsevier.de/ijleo Skin layer detection of optical coherence tomography images Mohammad R.N. Avanaki a,* , Ali Hojjatoleslami b a Research and Development Centre, School of Biosciences, University of Kent, Canterbury, Kent CT2 7PD, United Kingdom b Applied Optics Group (AOG), School of Physical Sciences, University of Kent, Canterbury, Kent CT2 7NH, United Kingdom a r t i c l e i n f o Article history: Received 8 November 2012 Accepted 2 April 2013 Keywords: Optical coherence tomography (OCT) Swept source OCT Skin layer detection a b s t r a c t In this paper, we present an image processing algorithm to automatically and more precisely detect the boundary between the main skin layers: stratum corneum, epidermis, and dermis. The aim of the proposed skin layer detection algorithm is to assist the dermatologists to measure the epidermal thickness (ET) for skin diseases diagnosis and also to assist pharmacologists so that they can make a better decision for prescribing according to the advancement of the skin disorders characterized with ET change. © 2013 Elsevier GmbH. All rights reserved. 1. Introduction For epidermal thickness (ET) measurement, skin layers are detected. Then the distance between stratum corneum and epi- dermis is measured [1]. ET measurement has been reported in Ref. [2–5], demonstrating the dependence between the health of skin and its physiological changes. Quantification of treatment effects is possible, for instance by monitoring the ET following the swelling of the horny layer due to the application of a moisturizer [2]. Mea- surement of the ET has also been studied as a possible procedure to be used in photodynamic therapy [6], in transepidermal drug delivery and in understanding the cutaneous reactions [7,8]. For instance, inflammatory skin diseases lead to thickening of the epi- dermis [2] and for that the ET measurement can be a useful method of diagnosis. Skin layers are usually characterized by identifying the first and second intensity peaks of the A-line in the B-scan OCT image corre- sponding to the top of the uppermost papillae and the valleys of the papillae [2,6,9–14]. In a number of studies traditional manual skin layers detection has been employed. In some other studies a semi- automatic border detection algorithm has been used in which user defined thresholds were used. Blomgren et al. proposed a model of skin and an automatic algorithm to detect the skin borders [9]. In this model different structural types of dermal papillae have been taken into account. The algorithm in Ref. [11] finds the peaks in the A-lines and only the peaks that can pass some statistical thresholds are chosen. In most of these approaches, the algorithm is dependent on the border markers which are defined manually. After detecting * Corresponding author. E-mail address: mrn.avanaki@seas.wustl.edu (M.R.N. Avanaki). the peaks on each A-line, they are joined using methods such as rubber-band [15] or interpolation by polynomial fitting [11]. In this study we utilize a set of image processing techniques to implement an algorithm to automatically detect the skin layers in OCT B-scan image regardless of the epidermal–dermal layer archi- tecture which has been a source of error in many automatic skin layers detection algorithms. The algorithm applied on 31 fingertip OCT images of variety of ethnic groups and found the layers suc- cessfully. The ET measurements have not been discussed in this paper. 2. Materials and method 2.1. OCT system configuration In this study, a swept-source Fourier domain OCT (SS-OCT) from Michelson Diagnostic TM has been employed for imaging. The light source of the OCT is a swept source works with central wavelength of 1305 nm and laser wavelength sweep range of 150 nm. The pen- etration depth of the system was measured as 1.5 mm. The OCT is based on multi beam technology in which the depth of focus is composed of four consecutive confocal gates each of 0.25 mm. Uti- lizing the multi beam technology, the images obtained from the four channels are averaged to generate images with a higher sig- nal to noise ratio (SNR) and contrast than those obtained from a conventional OCT instrument. The images obtained from this OCT system are B-scan OCT images with the lateral and axial resolutions of 7.5 m and 10 m, respectively. 2.2. Image processing algorithm for skin layer detection Our proposed skin layers detection algorithm is based on find- ing the high gradient of the B-scan image obtained by the SS-OCT, 0030-4026/$ – see front matter © 2013 Elsevier GmbH. All rights reserved. http://dx.doi.org/10.1016/j.ijleo.2013.04.033