acia-workflows: Automated Single-cell Imaging Analysis for Scalable and Deep Learning-based Live-cell Imaging Analysis Workflows

Abstract

Live-cell imaging (LCI) technology enables the detailed spatio-temporal characterization of living cells at the single-cell level, which is critical for advancing research in the life sciences, from biomedical applications to bioprocessing. High-throughput setups with tens to hundreds of parallel cell cultivations offer the potential for robust and reproducible insights. However, these insights are obscured by the large amount of LCI data recorded per experiment. Recent advances in state-of-the-art deep learning methods for cell segmentation and tracking now enable the automated analysis of such large data volumes, offering unprecedented opportunities to systematically study single-cell dynamics. The next key challenge lies in integrating these powerful tools into accessible, flexible, and user-friendly workflows that support routine application in biological research. In this work, we present acia-workflows, a platform that combines three key components: (1) the Automated live-Cell Imaging Analysis (acia) Python library, which supports the modular design of image analysis pipelines offering eight deep learning segmentation and tracking approaches; (2) workflows that assemble the image analysis pipeline, its software dependencies, documentation, and visualizations into a single Jupyter Notebook, leading to accessible, reproducible and scalable analysis workflows; and (3) a collection of application workflows showcasing the analysis and customization capabilities in real-world applications. Specifically, we present three workflows to investigate various types of microfluidic LCI experiments ranging from growth rate comparisons to precise, minute-resolution quantitative analyses of individual dynamic cells responses to changing oxygen conditions. Our collection of more than ten application workflows is open source and publicly available at https://github.com/JuBiotech/acia-workflows.

Figures (6)

• Figure 1: Five-step MLCI analysis pipeline (A), implemented in modular components of the acia library (B) utilizing SOTA methods and existing Python libraries. These steps are implemented sequentially within a single Jupyter Notebook, fusing code, documentation, software dependencies, and visualizations into a single workflow (C). This workflow is automatically scaled to high-throughput experiments with numerous time-lapse recordings to gain quantitative insights (D). Our workflow collection (E) showcases the importance of the six key capabilities: accessibility, customizability, modularity, scalability, shareability, and reproducibility (ACMS2R).
• Figure 2: Data extraction within acia-workflows.A: The 2D+t time-lapse is segmented to extract cell instances. B: Spatial single-cell features are extracted with units. C & D: Cell instances are associated through time to build the tracking and tracklet lineages. E: Temporal single-cell features are extracted with units and linked to the spatial information using the "label" key. All data is stored in common Python data structures of the NumPy, networkx, and Pandas libraries.
• Figure 3: Three variants to quantify colony growth rates: cell count (CC, upper row), total colony area (TCA, middle row), total single-cell area (TSCA, lower row) measured for C. glutamicum wildtype cultivated in BHI growth medium under constant flow conditions Seiffarth2025. The growth rates are inferred from time-series data (crosses) and fitted (dashed) using a log-linear growth model (right column).
• Figure 4: Co-culture cultivation with fluorescence-based strain labeling. Two C. glutamicum strains with E2-Crimson (red) and mVenus (blue) simultaneously cultivated in CGXII medium. (A) Snapshots of the time-lapse with cell outlines colored red and blue depending on their fluorescent label. (B) Temporal development of the measured TSCA of the two sub-populations (crosses) and the fit of an exponential growth model (dashed). (C-D) Average cell area distribution of the sub-populations, with a variance tube (one standard deviation) in light colors. (E) Average fluorescence of the labeled strains (blue/red) with variances (one standard deviation).
• Figure 5: Single-cell insights under oxygen switches. (A) TSCA development of E. coli as it has been reported by Kasahara et al.kasahara_unveiling_2025 at alternating aerobic (21% O2, green) and anaerobic (0% O2, red) cultivation conditions. (B) Single-cell area development and instantaneous growth rate (IGR) for five individual cells undergoing the switch from the aerobic (green) to anaerobic (red) cultivation conditions at $t=1.5~h$. (C) Lineage tree generated with automated cell tracking.