Active mesh and neural network pipeline for cell aggregate segmentation
Preprint
Abstract
Segmenting cells within cellular aggregates in 3D is a growing challenge in cell biology, due to improvements in capacity and accuracy of microscopy techniques. Here we describe a pipeline to segment images of cell aggregates in 3D. The pipeline combines neural network segmentations with active meshes. We apply our segmentation method to cultured mouse mammary duct organoids imaged over 24 hours with oblique plane microscopy, a high-throughput light-sheet fluorescence microscopy technique. We show that our method can also be applied to images of mouse embryonic stem cells imaged with a spinning disc microscope. We segment individual cells based on nuclei and cell membrane fluorescent markers, and track cells over time. We describe metrics to quantify the quality of the automated segmentation. Our segmentation pipeline involves a Fiji plugin which implement active meshes deformation and allows a user to create training data, automatically obtain segmentation meshes from original image data or neural network prediction, and manually curate segmentation data to identify and correct mistakes. Our active meshes-based approach facilitates segmentation postprocessing, correction, and integration with neural network prediction. Statement of significance In vitro culture of organ-like structures derived from stem cells, so-called organoids, allows to image tissue morphogenetic processes with high temporal and spatial resolution. Three-dimensional segmentation of cell shape in timelapse movies of these developing organoids is however a significant challenge. In this work, we propose an image analysis pipeline for cell aggregates that combines deep learning with active contour segmentations. This combination offers a flexible and efficient way to segment three-dimensional cell images, which we illustrate with by segmenting datasets of growing mammary gland organoids and mouse embryonic stem cells.Keywords
All Related Versions
- Published version: Biophysical Journal, 122 (9), 1586.
This publication has 34 references indexed in Scilit:
- Multiple cell and population-level interactions with mouse embryonic stem cell heterogeneityDevelopment, 2015
- Efficient Bayesian-based multiview deconvolutionNature Methods, 2014
- Optically sectioned imaging by oblique plane microscopyOptics Express, 2008
- Segmentation of touching cell nuclei using gradient flow trackingJournal of Microscopy, 2008
- Lifeact: a versatile marker to visualize F-actinNature Methods, 2008
- A global double‐fluorescent Cre reporter mouseGenesis, 2007
- Multi-view image fusion improves resolution in three-dimensional microscopyOptics Express, 2007
- Optical Sectioning Deep Inside Live Embryos by Selective Plane Illumination MicroscopyScience, 2004
- Induction of mammary tumors by expression of polyomavirus middle T oncogene: a transgenic mouse model for metastatic disease.Molecular and Cellular Biology, 1992
- Snakes: Active contour modelsInternational Journal of Computer Vision, 1988