Introduction. Assessing metastatic potential (MP) based on morphological parameters of cancer cells is a promising direction in experimental oncology. The shape of a cell often correlates with an invasive phenotype, characterized by greater elongation and decreased roundness. A deep neural network image-segmentation algorithm implemented in Omnipose lacks tools for obtaining morphological data and their subsequent analysis for MP evaluation. Our aim was to develop a specialized software module for complete morphometric analysis of cells in microscopic images without need in specific pre-processing or staining.

Materials and Methods. Custom Python code (used packages: matplotlib.pyplot, skimage, scipy, google.colab, cellpose) was integrated into a pre-trained Omnipose model. The created module performs end-to-end analytics and calculation of key morphologic parameters: Area, Perimeter, Major_axis, Minor_axis, Circularity, Aspect_Ratio, Roundness, Solidity. The main innovation is an algorithm that analyzes these parameters, assigns characteristics to each cell, and outputs averaged values for further assessment of MP. Segmentation performance was evaluated by comparison with manual analysis using microscopic images of 65 breast cancer cells with high (MDA-MB-231) and low (MCF-7) MP.

Results. The Omnipose neural network's cell recognition accuracy exceeded 80%, confirming the high reliability of the approach. Measured morphologic parameters: Area, Perimeter, Major_axis, Minor_axis demonstrated strong agreement with manual analysis, with errors below 5%. The shape descriptors: Circularity, Solidity, Roundness and Aspect_Ratio demonstrated high concordance, with a maximum deviation of 11% from manually calculated values.

Conclusions. This work developed an analytical module that transforms Omnipose from a segmentation tool into a platform for direct morphometric analysis of microscopic images. The module serves as a foundation for a convenient tool to quantitatively evaluate cancer cell metastasis.