This project utilizes advanced machine learning models to segment Salmonella enterica serovar Typhimurium in complex, time-lapse microscopy image series, where primary human epithelial cell monolayers are infected with wild-type Salmonella. Using a recently developed unique imaging platform developed by our group, we capture high-resolution, multi-dimensional data of host-pathogen interactions in real time. The dynamic nature of these datasets, with significant cell movement and morphological changes, presents challenges for traditional analysis. Due to the limitations of fluorescence microscopy, such as phototoxicity and limited temporal resolution, we employ Differential Interference Contrast (DIC) microscopy, which offers the ideal approach for high-resolution imaging with minimal impact on cell viability. While U-net architectures have shown promise in segmenting these images (BactUnet_V1), we are exploring other deep learning models, such as attention-based networks and recurrent neural networks (RNNs), to improve robustness and accurately capture both bacterial and host cell responses. By integrating advanced microscopy with machine learning, we aim to set a new standard for automated analysis of infection dynamics, providing novel insights into host-pathogen interactions and to advance the field of infection biology.