This research addresses the important task of developing an effective method for automatic brain tumor detection based on magnetic resonance imaging (MRI) images. This is a key task in medical diagnostics, as it can significantly improve the early detection and treatment of the disease. This paper uses an approach that combines deep learning and convolutional neural networks (СNNs) and uses parallelization to optimize computational processes and reduce processing time. This approach proves to be effective in diagnosing brain cancer, leading to an increase in system efficiency. In this paper, we propose an algorithm that uses Message Passing Interface (MPI) and Compute Unified Device Architecture (CUDA) Python technologies for parallelized learning and tumor detection in MRI images. As a result of the experiments, a significant acceleration was achieved, which is about 3 times for MPI technology and about 24 times for CUDA, without compromising the diagnostic accuracy compared to previous studies by domestic and foreign scientists. The proposed approach is promising for the implementation of automated systems in medical institutions, providing accurate results of brain cancer detection in real-time. Further testing and evaluation of the algorithm on computing systems with more cores and on larger datasets is recommended to better understand the ratio of accuracy and computing time.