Abstract:
This paper is devoted to topical issues - the development of methods for analyzing texture images of breast cancer. The main problem that
is resolved in the article is that the requirements for the results of pre-processing are increasing. As a result of the task, images of magnetic resonance
imaging of the breast are considered for image processing using texture image analysis methods. The main goal of the research is the development
and implementation of algorithms that allow detecting and isolating a tumor in the breast in women in an image. To solve the problem, textural features,
clustering, orthogonal transformations are used. The methods of analysis of texture images of breast cancer, carried out in the article, namely: Hadamard
transform, oblique transform, discrete cosine transform, Daubechies transform, Legendre transform, the results of their software implementation
on the example of biomedical images of oncological pathologies on the example of breast cancer, it is shown that The most informative for image
segmentation is the method based on the Hadamard transform and the method based on the Haar transform. The article presents recommendations
for using the results in practice, namely, it is shown that clinically important indicators that make a significant contribution to assessing the degree
of pathology and the likelihood of developing diseases, there are other information parameters: diameter, curvature, etc. Therefore, increased
requirements for the reliability, accuracy, speed of processing biomedical images.
