Neural network classification system for pigmented skin neoplasms with preliminary hair removal in photographs

Abstract

The article proposes a neural network classification system for pigmented skin neoplasms with a preliminary processing stage to remove hair from the images. The main difference of the proposed system is the use of the stage of preliminary image processing to identify the location of the hair and their further removal. This stage allows you to prepare dermatoscopic images for further analysis in order to carry out automated classification and diagnosis of pigmented skin lesions. Modeling was carried out using the MatLAB R2020b software package on clinical dermatoscopic images from the international open archive ISIC Melanoma Project. The proposed system made it possible to increase the recognition accuracy of pigmented skin lesion images in 10 diagnostically important categories up to 80.81%. The use of the proposed system for the recognition and classification of images of dermatoscopic pigmented lesions by specialists will make it possible to increase the diagnostic efficiency in comparison with methods of visual diagnosis, and will also allow starting treatment at an earlier stage of the disease, which directly affects the survival and recovery rates for patients.