Neural-network-based classification of skin structural elements in case of urticaria using histological RGB images

We address the problem of differential diagnostics of chronic spontaneous urticaria and urticarial vasculitis. The clinical pictures of these two allergy diagnoses are similar, however a well-trained pathologist can see the differences in skin tissue on the microscopic level, and thus the histological study of skin biopsy is usually performed to prescribe the right treatment. To increase the throughput and quality of the histological study and differential diagnostics of chronic spontaneous urticaria and urticarial vasculitis, we propose an imaging system with neural-network-based classification of skin tissue structures. The capabilities of a hyperspectral microscopic visualization system with acousto-optical module for increasing the efficiency of neural network training are also being considered.