APPLICATION OF DEEP LEARNING NEURAL NETWORK ALGORITHMS FOR BRAIN TUMOR CLASSIFICATION 

Maksim O. Timoshkin, Student, Moscow State University named after M.V. Lomonosov (1 Leninskie Gory street, Moscow, Russia), E-mail: max.timoshkin@inbox.ru
Elena G. Romanova, Candidate of technical sciences, associate professor, associate professor of the sub-department of higher and applied mathematics, Penza State University (40 Krasnaya street, Penza, Russia), E-mail: romanova.elenar2016@yandex.ru 

Background. Deep learning is a rapidly developing area of machine learning that allows you to find dependencies in semi-structured data. The relevance of the work lies in the fact that currently automatic classification of tissue types plays an important role in computer diagnostics. The goal of the work is to automate the process of determining the type of brain tumor from its image using deep learning methods. Materials and methods. The work uses deep learning methods to automatically classify the type of brain tumor based on its image. Results and conclusions. A comparison was made of neural network models with different architectures, in different training modes, and with and without sample enrichment. During the training process, using the best architecture, it was possible to achieve a quality of 96 % on the validation set. 

brain tumor classification, deep learning, neural network architectures, sample enrichment, MRI images, learning modes 

Timoshkin M.O., Romanova E.G. Application of deep learning neural network algorithms for brain tumor classification. Nadezhnost' i kachestvo slozhnykh sistem = Reliability and quality of complex systems. 2024;(3):51–65. (In Russ.). doi: 10.21685/2307-4205-2024-3-6