Srivastava, R., Yow, A. P., Cheng, J., Wong, D. W., & Tey, H. L. (2017, August). Supervised 3D graph-based automated epidermal thickness estimation. In Signal and Image Processing (ICSIP), 2017 IEEE 2nd International Conference on (pp. 297-301).
Abstract:
Skin biopsies are frequently performed for the diagnosis of skin diseases. However, the invasive nature of biopsies confers many disadvantages. Non-invasive imaging using optical coherence tomography (OCT) with automated disease detection can help reduce requirements for biopsies. One of the features in skin diseases is abnormal epidermal thickness. Automated determination of epidermal thickness requires segmenting the epidermis and this paper presents a novel method of supervised graph-based epidermis segmentation. The cost function for the segmentation method is learned from manually marked images. Epidermis segmentation is followed by epidermal thickness estimation. Evaluation of the method on a dataset containing 10 OCT volumes gives promising results which demonstrate the utility of the proposed method for epidermis segmentation and epidermal thickness measurement.