Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

EAP4EMSIG -- Experiment Automation Pipeline for Event-Driven Microscopy to Smart Microfluidic Single-Cells Analysis

Nils Friederich, Angelo Jovin Yamachui Sitcheu, Annika Nassal, Matthias Pesch, Erenus Yildiz, Maximilian Beichter, Lukas Scholtes, Bahar Akbaba, Thomas Lautenschlager, Oliver Neumann, Dietrich Kohlheyer, Hanno Scharr, Johannes Seiffarth, Katharina Nöh, Ralf Mikut

TL;DR

EAP4EMSIG presents a modular eight-module pipeline for automated, event-driven microscopy in microfluidic single-cell analysis, integrating microscope control, real-time processing, data management in OMERO, and AI-assisted annotation. Zero-shot segmentation comparisons on dataMicrobsTrackingMillion show Omnipose delivering the highest PQ (0.9336) while Contour Proposal Network offers the fastest inference (185 ms) with strong performance, whereas Segment Anything underperforms for microbial shapes. The work demonstrates the feasibility of real-time analytics and planning to guide experiments, with future work focusing on domain-specific retraining and achieving real-time latency targets around 100 ms. The approach promises scalable, automated experimentation in microbial microfluidics, enabling higher throughput and more consistent event-driven decision-making in bioscience workflows.

Abstract

Microfluidic Live-Cell Imaging (MLCI) generates high-quality data that allows biotechnologists to study cellular growth dynamics in detail. However, obtaining these continuous data over extended periods is challenging, particularly in achieving accurate and consistent real-time event classification at the intersection of imaging and stochastic biology. To address this issue, we introduce the Experiment Automation Pipeline for Event-Driven Microscopy to Smart Microfluidic Single-Cells Analysis (EAP4EMSIG). In particular, we present initial zero-shot results from the real-time segmentation module of our approach. Our findings indicate that among four State-Of-The- Art (SOTA) segmentation methods evaluated, Omnipose delivers the highest Panoptic Quality (PQ) score of 0.9336, while Contour Proposal Network (CPN) achieves the fastest inference time of 185 ms with the second-highest PQ score of 0.8575. Furthermore, we observed that the vision foundation model Segment Anything is unsuitable for this particular use case.

EAP4EMSIG -- Experiment Automation Pipeline for Event-Driven Microscopy to Smart Microfluidic Single-Cells Analysis

TL;DR

EAP4EMSIG presents a modular eight-module pipeline for automated, event-driven microscopy in microfluidic single-cell analysis, integrating microscope control, real-time processing, data management in OMERO, and AI-assisted annotation. Zero-shot segmentation comparisons on dataMicrobsTrackingMillion show Omnipose delivering the highest PQ (0.9336) while Contour Proposal Network offers the fastest inference (185 ms) with strong performance, whereas Segment Anything underperforms for microbial shapes. The work demonstrates the feasibility of real-time analytics and planning to guide experiments, with future work focusing on domain-specific retraining and achieving real-time latency targets around 100 ms. The approach promises scalable, automated experimentation in microbial microfluidics, enabling higher throughput and more consistent event-driven decision-making in bioscience workflows.

Abstract

Microfluidic Live-Cell Imaging (MLCI) generates high-quality data that allows biotechnologists to study cellular growth dynamics in detail. However, obtaining these continuous data over extended periods is challenging, particularly in achieving accurate and consistent real-time event classification at the intersection of imaging and stochastic biology. To address this issue, we introduce the Experiment Automation Pipeline for Event-Driven Microscopy to Smart Microfluidic Single-Cells Analysis (EAP4EMSIG). In particular, we present initial zero-shot results from the real-time segmentation module of our approach. Our findings indicate that among four State-Of-The- Art (SOTA) segmentation methods evaluated, Omnipose delivers the highest Panoptic Quality (PQ) score of 0.9336, while Contour Proposal Network (CPN) achieves the fastest inference time of 185 ms with the second-highest PQ score of 0.8575. Furthermore, we observed that the vision foundation model Segment Anything is unsuitable for this particular use case.

Paper Structure

This paper contains 31 sections, 2 figures, 1 table.

Table of Contents

  1. Introduction
  2. What are microbes?
  3. Why are microbes relevant?
  4. Why is research on microbes essential?
  5. Why is the segmentation of microbes relevant?
  6. What is the challenge in microbe research?
  7. How does this paper address this challenge?
  8. Related Work
  9. Experiment Automation Pipelines.
  10. Segmentation.
  11. Experiment Database.
  12. Semi-Automated Data Annotations.
  13. AI-ready Image with Ground Truth Cell Simulation.
  14. Module Interaction.
  15. Methodology
  16. ...and 16 more sections

Figures (2)

  • Figure 1: EAP4EMSIG visualization. The pipeline consists of eight modules, represented by the light blue boxes and the database, arranged in a cyclical process. The microbial images in the figure come from dataset dataMicrobsTrackingMillion. The images from the experiment chip are from an internal dataset.
  • Figure 2: Comparison of zero-shot instance segmentation predictions for dataMicrobsTrackingMillion. The original image is shown in \ref{['fig:results_org']} and the predictions are shown in \ref{['fig:results_omnipose']} to \ref{['fig:results_sam']}.