The changes in the quantitative parameters and spatial structure of collagen are considered a key diagnostic and prognostic factor associated with the development of many malignant neoplasms, including breast cancer (BCa). The aim of the work was to develop and test an algorithm for the assessment of collagen organization parameters as informative attributes associated with BCa for developing technology of machine learning and building an intelligent system of cancer diagnostics.
Materials and methods: Tumor tissue samples of 5 patients with breast fibroadenomas and 20 patients with stage I-II BCa were studied. Collagen was identified histochemically by Mallory method. Photomicrographs of the studied preparations were obtained using a digital microscopy complex AxioScope A1. Morphometric studies were performed using the software CurveAlign v. 4.0. beta and ImageJ.
Results: The algorithm for determining the quantitative characteristics and spatial organization of the collagen matrix in tumor tissue samples has been developed and tested. We showed that collagen fibers in the BCa tissue are characterized by significantly lower values of length (p < 0.001) and width (p < 0.001) as well as higher values of straightness (p < 0.001) and angle (p < 0.05) compared to these in the fibroadenoma tissue. No significant difference was found in the density of collagen fibers in the tissue of benign and malignant neoplasms of the mammary gland.
Conclusion: The algorithm allows assessing a wide range of parameters of collagen fibers in tumor tissue, including their spatial orientation and mutual arrangement, parametric characteristics and density of the three-dimensional fibrillar network.