Vytvorenie aplikácie na detekciu buniek v mikroskopických snímkach

Abstract

Cell detection and segmentation in microscopic images play a key role in biomedical diagnostics and research. This bachelor thesis focuses on the development of a user-friendly application enabling automated cell analysis using modern deep learning methods. The core of the solution lies in the integration of segmentation models based on convolutional neural networks, implemented using the PyTorch and Keras libraries. The application supports training of custom models, segmentation of new images, and extraction of quantitative cell parameters for further analysis. Emphasis is placed on ease of use without the need for advanced programming skills, ensuring that the resulting tool streamlines the processing of microscopic data, enhances the reproducibility of results, and enables faster interpretation of biological changes by experts. The application can be used in clinical practice to support more accurate diagnosis, as well as in research settings for time-lapse experiment analysis. This work thus contributes to the accessibility of advanced segmentation methods for a broader community of users in the biomedical field.