This paper examines the problem of automatically detecting and classifying brain tumors based on magnetic resonance imaging (MRI) of the brain. The study analyzes the existing shortcomings of traditional machine learning and deep learning approaches, in particular their limitation to local features and their insufficient consideration of global spatial dependencies. To address this problem, a hybrid model combining convolutional neural networks with the self-attention mechanism of the Transformer architecture is proposed. The model enables the joint learning of local and global image features. Experiments were conducted on a four-class (glioma, meningioma, no tumor, and pituitary tumor) MRI image dataset. The results demonstrated the stable performance of the proposed model, medium-to-high accuracy, and good generalization ability. The computational complexity and stability of the model were also theoretically substantiated.
| Mualliflar | D. T. Muhamediyeva, N.S. Mamatov, A.F. Isoqov, I.J. Juraev, Sh.E. Jovliboyev |
|---|---|
| Jurnal | Samarqand davlat universiteti ilmiy tadqiqotlar axborotnomasi |
| Nashr sanasi | 2026-07-01 |
| Jild | 2 |
| Son | 3 |
| Betlar | 186-194 |
| DOI | 10.59251/2181-3973.2025.v1.138.1.3994 |
DOI: 10.59251/2181-3973.2025.v1.138.1.3994 · Maqolaning asl sahifasi
We consider the problem of finding a solution of the mixed problem for the negative order modified Korteweg-de Vries–Liouville equation with a self-consistent integral source using the inverse spectral problem for a…
This paper applies the inverse spectral problem method to integrate the Hirota equation with an additional term and a finite density within the class of periodic infinite-gap functions. The evolution of the spectral…
In this study, Fe₃O₄ and CoFe₂O₄ magnetic nanostructures were successfully synthesized by a modified coprecipitation method with the participation of a natural stabilizer, and their structural, electronic, and magnetic…
BaTiO₃-based perovskite materials were synthesized and comprehensively characterized by energy-dispersive X-ray spectroscopy (EDS), Fourier transform infrared spectroscopy (FTIR), and UV–Vis absorption techniques. EDS…
Activated carbons were synthesized from walnut shells by pyrolysis and chemical activation, and their structural and adsorption properties were studied. BET analysis showed that the surface area increased from 285.4…
The accelerating urbanization of the Fergana Valley has produced one of the densest settlement systems in Central Asia, yet the agglomerations that anchor this system are habitually examined within national borders, as…
This paper is devoted to the problem of finding a local parametric representation of solutions to a nonlinear algebraic system in the neighborhood of singular points. At points where the Jacobian matrix becomes…
This study investigates the specific activities of natural (226Ra, 232Th, 40K) and technogenic (137Cs) radionuclides in soils and food products from selected regions of Uzbekistan using scintillation gamma spectrometry…
A methodology has been developed for optimizing the identification, recognition, and classification of micro-objects using redundant image information structures - histological, morphological, fractal, correlation, and…
Addressing the challenges of ambiguity in veterinary medicine, this study proposes a novel approach to cattle disease diagnosis using neutrosophic sets and the Sugeno fuzzy model. The research develops specialized…
Samarqand davlat universiteti ilmiy tadqiqotlar axborotnomasi — barcha maqolalar