OrganoidTracker: Efficient cell tracking using machine learning and manual error correction.

Rutger N U Kok,Xuan Zheng,Sander J Tans,Katarzyna Bozek,Jeroen S Van Zon,Guizela Huelsz-Prince,Yvonne J Goos,Laetitia Hebert,Greg J Stephens,Joseph Najbauer

doi:10.1371/journal.pone.0240802

Abstract

Time-lapse microscopy is routinely used to follow cells within organoids, allowing direct study of division and differentiation patterns. There is an increasing interest in cell tracking in organoids, which makes it possible to study their growth and homeostasis at the single-cell level. As tracking these cells by hand is prohibitively time consuming, automation using a computer program is required. Unfortunately, organoids have a high cell density and fast cell movement, which makes automated cell tracking difficult. In this work, a semi-automated cell tracker has been developed. To detect the nuclei, we use a machine learning approach based on a convolutional neural network. To form cell trajectories, we link detections at different time points together using a min-cost flow solver. The tracker raises warnings for situations with likely errors. Rapid changes in nucleus volume and position are reported for manual review, as well as cases where nuclei divide, appear and disappear. When the warning system is adjusted such that virtually error-free lineage trees can be obtained, still less than 2% of all detected nuclei positions are marked for manual analysis. This provides an enormous speed boost over manual cell tracking, while still providing tracking data of the same quality as manual tracking.

Highlights

Scientists are looking at development and homeostasis of organs at the single-cell level [1,2,3]
The approach was the same as for the F1 score: any detected nucleus center more than 5, m away from a nucleus center in the manual tracking data was deleted
We have developed a program for cell tracking based on nuclear markers inside organoids that uses a convolutional neural network for detection and a min-cost flow solver for linking cell positions

Summary

Introduction

Scientists are looking at development and homeostasis of organs at the single-cell level [1,2,3]. In our cell tracking approach, we use a convolutional neural network for cell detection, a min-cost flow solver [23] for linking cells over time and a small set of rules to detect improbable cell tracking data that the user needs to verify and correct. The nucleus center positions from our manually annotated tracking dataset were used to train a fully convolutional neural network, where the input and output are 3D images of the same shape.

Results

Conclusion