Development of a comprehensive method for the dermatoscopic images analysis of the facial skin with acne

Abstract

Background: One of the most common inflammatory chronic and recurrent skin diseases is acne (“acne vulgaris”), which appears itself as open or closed comedones and inflammatory skin lesions in the form of papules, pustules, nodes, etc. It has been established that acne is one of the most common dermatoses, since, according to modern data, it affects about 9.4% of the population. During adolescence, up to 90% of people suffer, and in adulthood — about 20% with varying degrees of severity. Currently, there are many approaches to treating this disease through various cosmetic treatments such as phototherapy, ultrasonic skin cleansing, Mesotherapy, chemical peels, and medication. Therefore, the development of methods and means of differential diagnosis of acne is one of the urgent tasks in the field of biomedical engineering, dermatology, and clinical medicine, since this allows timely identification of the localization of the disease, its causes, and prescribing appropriate treatment. However, the solution to the problem of monitoring the dynamics of external manifestations of the disease is possible only with the use of combined mathematical methods for image analysis. Objectives: To develop a comprehensive method for analyzing dermatoscopic images for monitoring the external manifestations of acne disease during treatment and isolating the affected areas of the facial skin. Materials and Methods: Dermatological preclinical researches of the skin were conducted in the laboratory of 3D-biomedical technologies of the Department of Biomedical Engineering of the Kharkiv National University of Radio Electronics, using a digital videodermatoscope BIO Bm6+ in daylight and a portable skin analyzer Skin Scope F-102 in the ultraviolet range. Clinical researches were conducted based on the Department of Pediatric Propaedeutics #2 of the Kharkiv National Medical University. The development of a software tool for image analysis was conducted out in Python programming using the libraries OpenCV, Scikit-image, Numpy, PIL, Mathplotlib. Determination of the affected skin areas and calculation of the parameters of inflammation were carried out using multi-Otsu methods and morphological segmentation of digital dermatoscopic images. Results: During the research, automated software was developed that allows to analyze in dynamics the nature of inflammatory processes and the area of facial skin lesions, as well as to carry out a differential diagnosis of acne disease. The proposed method for the analysis of dermatoscopic images makes it possible to perform color segmentation and obtain a map of the gradations of skin inflammations to control the dynamics during the prescribed treatment. Conclusions: The comprehensive method of analysis of dermatoscopic images of the skin of the face makes it possible to effectively control the condition of the skin of the face from acne during treatment, while analyzing the degree of inflammatory processes and the area of lesions, where, using the developed software, in an automated mode, red gradations are calculated to detect the boundaries of inflammation, geometric parameters and percentage of lesions in relation to healthy facial skin.