UC Santa Cruz

Understanding the dynamic responses of cellular processes to stimuli is crucial for uncovering the regulatory mechanisms underlying complex biological phenomena. This work primarily focuses on modeling and analyzing biological system responses over time. In some cases, there is no stimulus, while in others, stimuli such as electric fields (EF), nutrient exposure, or initial wounding are present. This study underscores the significance of considering dynamics in biological systems and the need to develop methods for analyzing and modeling temporal dynamics.

In the first part of our research, we focus on macrophages, a type of immune cell, and their subtypes (M0, M1, and M2). We develop robust image processing methods for single-cell segmentation and tracking using single-cell time-lapse microscopy and label-free live-cell imaging. By mapping the morphological features to cell migratory behavior, we train a deep-learning model to classify macrophage subtypes. Our findings reveal distinct migratory behaviors for M1 and M2 macrophages, demonstrating that cell motility and morphology can effectively identify functionally diverse macrophage phenotypes. This has significant implications for developing cost-efficient, high-throughput screening methods targeting macrophage polarization. Building on this foundation, the second part of our work explores cell subtype classification in the context of galvanotaxis, where cells migrate in response to electric fields (EF). We extend our image processing and machine learning framework to control and analyze cell migration under EF, providing new insights into the mechanisms driving cellular responses to electrical stimuli. In the final segment of this thesis, we apply our single-cell microscopy image processing techniques to study bacterial spore germination, developing a predictive model for spore germination with various germinants administered at different intervals. This interdisciplinary approach enhances our understanding of spore biology and demonstrates the versatility and applicability of our image-processing methods across different biological systems.

This study presents a comprehensive approach that combines advanced image processing techniques, machine learning algorithms, and experimental methodologies to model and analyze biological system responses over time. Additionally, we highlight the importance of non-computationally expensive methods for processing images and quantifying behavior, enabling real-time control and analysis.