h1

TY  - THES
AU  - Ruzaeva, Karina
TI  - Geometry-aware image analysis for microfluidic live-cell experimentation
PB  - RWTH Aachen University
VL  - Dissertation
CY  - Aachen
M1  - RWTH-2025-03601
SP  - 1 Online-Ressource : Illustrationen
PY  - 2025
N1  - Veröffentlicht auf dem Publikationsserver der RWTH Aachen University
N1  - Dissertation, RWTH Aachen University, 2025
AB  - Time-lapse microscopy combined with advanced imaging techniques offers new opportunities to approach fundamental and applied biological questions. Specifically, microfluidic tools enable the study of living cells under precisely controlled environmental conditions (medium, temperature, light) and allow for their observation through live-cell microscopy. This approach enables the capture of time-resolved cellular responses, which are recorded as high-volume time-lapse image sequences. Processing these large-scale image sequences poses numerous challenges for image analysis, such as varying noise levels, low-intensity gradients and limited image capture rates. Addressing these challenges is crucial for developing biotechnological processes as it becomes increasingly necessary better to understand the behavior of microorganisms at the single-cell level. Advanced image-processing techniques play a critical role in extracting valuable insights from these datasets, which are essential for optimizing biotechnological applications. The dissertation “Geometry-aware image analysis for microfluidic live-cell experimentation” emphasizes the importance of incorporating prior knowledge of microorganisms' geometry and behavior—specifically their shape, size, and division mechanisms—into image analysis techniques. In this dissertation, we present an image processing workflow comprising ground truth data generation, segmentation, and tracking. This ground truth generation and subsequent segmentation and tracking algorithms are informed by the geometry and behavioral characteristics of cells, which are determined by the selected microorganism. To implement this workflow, this dissertation proposes ground truth data generation methods that include both synthetic image simulations and the processing of annotations from real data. For the segmentation task, we combine geometry-aware variational spline-based segmentation with machine learning-based detection to enhance the accuracy of cell identification. This approach is complemented by activity-based tracking that monitors cell behavior over time, enabling the extraction of critical parameters such as cell size, count, and dynamic behavior. The extracted data can be used to dynamically adjust bioprocess conditions to optimize growth and yield, leading to greater efficiency and productivity in biotechnological processes. By integrating this geometry and behavior-aware image processing methods, including ground truth generation, geometry-aware segmentation, and activity-based tracking, this dissertation underscores the potential of precise image analysis in enhancing live-cell experimentation. The results not only improve the accuracy of single-cell analysis but also help to optimize biotechnological processes by understanding cellular behavior.
LB  - PUB:(DE-HGF)11
DO  - DOI:10.18154/RWTH-2025-03601
UR  - https://publications.rwth-aachen.de/record/1009666
ER  -