Semi-automated Algorithm for Localization of Dermal/ Epidermal Junction in Reflectance Confocal Microscopy Images of Human Skin

Artificial Intelligence-Based Approaches to Reflectance Confocal Microscopy Image Analysis in Dermatology

Reflectance confocal microscopy (RCM) is a non-invasive imaging method designed to identify various skin diseases. Confocal based diagnosis may be subjective due to the learning curve of the method, the scarcity of training programs available for RCM, and the lack of clearly defined diagnostic criteria for all skin conditions. Given that in vivo RCM is becoming more widely used in dermatology, numerous deep learning technologies have been developed in recent years to provide a more objective approach to RCM image analysis. Machine learning-based algorithms are used in RCM image quality assessment to reduce the number of artifacts the operator has to view, shorten evaluation times, and decrease the number of patient visits to the clinic. However, the current visual method for identifying the dermal-epidermal junction (DEJ) in RCM images is subjective, and there is a lot of variation. The delineation of DEJ on RCM images could be automated through artificial intelligence, saving time and assisting novice RCM users in studying the key DEJ morphological structure. The purpose of this paper is to supply a current summary of machine learning and artificial intelligence’s impact on the quality control of RCM images, key morphological structures identification, and detection of different skin lesion types on static RCM images.

Wavelet-based statistical classification of skin images acquired with reflectance confocal microscopy

Detecting skin lentigo in reflectance confocal microscopy images is an important and challenging problem. This imaging modality has not yet been widely investigated for this problem and there are a few automatic processing techniques. They are mostly based on machine learning approaches and rely on numerous classical image features that lead to high computational costs given the very large resolution of these images. This paper presents a detection method with very low computational complexity that is able to identify the skin depth at which the lentigo can be detected. The proposed method performs multiresolution decomposition of the image obtained at each skin depth. The distribution of image pixels at a given depth can be approximated accurately by a generalized Gaussian distribution whose parameters depend on the decomposition scale, resulting in a very-low-dimension parameter space. SVM classifiers are then investigated to classify the scale parameter of this distribution allowing real-time detection of lentigo. The method is applied to 45 healthy and lentigo patients from a clinical study, where sensitivity of 81.4% and specificity of 83.3% are achieved. Our results show that lentigo is identifiable at depths between 50μm and 60μm, corresponding to the average location of the the dermoepidermal junction. This result is in agreement with the clinical practices that characterize the lentigo by assessing the disorganization of the dermoepidermal junction.

Remodeling of the Epithelial-Connective Tissue Interface in Oral Epithelial Dysplasia as Visualized by Noninvasive 3D Imaging

Early neoplastic features in oral epithelial dysplasia are first evident at the basal epithelium positioned at the epithelial-connective tissue interface (ECTI), separating the basal epithelium from the underlying lamina propria. The ECTI undergoes significant deformation in early neoplasia due to focal epithelial expansion and proteolytic remodeling of the lamina propria, but few studies have examined these changes. In the present study, we quantitated alterations in ECTI topography in dysplasia using in vivo volumetric multiphoton autofluorescence microscopy and second harmonic generation microscopy. The label-free method allows direct noninvasive visualization of the ECTI surface without perturbing the epithelium. An image-based parameter, "ECTI contour," is described that indicates deformation of the ECTI surface. ECTI contour was higher in dysplasia than control or inflamed specimens, indicating transition from flat to a deformed surface. Cellular parameters of nuclear area, nuclear density, coefficient of variation in nuclear area in the basal epithelium and collagen density in areas adjacent to ECTI were measured. ECTI contour differentiated dysplasia from control/benign mucosa with higher sensitivity and specificity than basal nuclear density or basal nuclear area, comparable with coefficient of variation in nuclear area and collagen density. The presented method offers a unique opportunity to study ECTI in intact mucosa with simultaneous assessment of cellular and extracellular matrix features, expanding opportunities for studies of early neoplastic events near this critical interface and potentially leading to development of new approaches for detecting neoplasia in vivo Cancer Res; 76(16); 4637-47. ©2016 AACR.

Validation Study of Automated Dermal/Epidermal Junction Localization Algorithm in Reflectance Confocal Microscopy Images of Skin

Reflectance confocal microscopy (RCM) has seen increasing clinical application for noninvasive diagnosis of skin cancer. Identifying the location of the dermal-epidermal junction (DEJ) in the image stacks is key for effective clinical imaging. For example, one clinical imaging procedure acquires a dense stack of 0.5×0.5mm FOV images and then, after manual determination of DEJ depth, collects a 5×5mm mosaic at that depth for diagnosis. However, especially in lightly pigmented skin, RCM images have low contrast at the DEJ which makes repeatable, objective visual identification challenging. We have previously published proof of concept for an automated algorithm for DEJ detection in both highly- and lightly-pigmented skin types based on sequential feature segmentation and classification. In lightly-pigmented skin the change of skin texture with depth was detected by the algorithm and used to locate the DEJ. Here we report on further validation of our algorithm on a more extensive collection of 24 image stacks (15 fair skin, 9 dark skin). We compare algorithm performance against classification by three clinical experts. We also evaluate inter-expert consistency among the experts. The average correlation across experts was 0.81 for lightly pigmented skin, indicating the difficulty of the problem. The algorithm achieved epidermis/dermis misclassification rates smaller than 10% (based on 25×25 mm tiles) and average distance from the expert labeled boundaries of ~6.4 μm for fair skin and ~5.3 μm for dark skin, well within average cell size and less than 2x the instrument resolution in the optical axis.