The study presents a software-based approach for automated segmentation of cerebral endothelial cells in bright-field images using the YOLOv11 deep-learning architecture. The training dataset annotations are generated from fluorescent (Hoechst) images processed with classical computer-vision methods. The developed algorithm enables the extraction of endothelial cell nuclei in brain microvessels and the exclusion of regions corresponding to macrophages, which reduces the number of false-positive detections. The resulting binary masks are used for automatic generation of annotations required for training the neural-network model. The proposed method is potentially applicable to all types of cells amenable to fluorescent staining and enables automation of data preparation and analysis of cell images obtained via bright-field microscopy, thereby expanding the possibilities for non-invasive research in optical biophotonics. Keywords: fluorescence microscopy, bright-field microscopy, machine learning, computer vision, image segmentation, endothelial cells, optical biophotonics.