RDCNet: Instance Segmentation with a Minimalist Recurrent Residual Network

Ortiz, Raphaël; Medeiros, Gustavo de; Peters, Antoine H.F.M.; Liberali, Prisca; Rempfler, Markus

doi:10.1007/978-3-030-59861-7_44

Cited by 7 publications

(5 citation statements)

References 20 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We used our ground-truth dataset to compare seven instance segmentation networks (in S1 Table ), including Cellpose [23], Stardist-3D [24], RDCNet [25], U3D-BCD [26], UNETR-BCD [27], ELEPHANT [28], and QCANet [6]. These methods span a variety of network architectures, from those including recurrent blocks or transformers to more conventional U-Nets.…”

Section: Benchmarking Of Seven Instance Segmentation Methods On Blast...mentioning

confidence: 99%

A novel ground truth dataset enables robust 3D nuclear instance segmentation in early mouse embryos

Nunley

Shao

Grover

et al. 2023

Preprint

View full text Add to dashboard Cite

For investigations into fate specification and cell rearrangements in live images of preimplantation embryos, automated and accurate 3D instance segmentation of nuclei is invaluable; however, the performance of segmentation methods is limited by the images' low signal-to-noise ratio and high voxel anisotropy and the nuclei's dense packing and variable shapes. Supervised machine learning approaches have the potential to radically improve segmentation accuracy but are hampered by a lack of fully annotated 3D data. In this work, we first establish a novel mouse line expressing near-infrared nuclear reporter H2B-miRFP720. H2B-miRFP720 is the longest wavelength nuclear reporter in mice and can be imaged simultaneously with other reporters with minimal overlap. We then generate a dataset, which we call BlastoSPIM, of 3D microscopy images of H2B-miRFP720-expressing embryos with ground truth for nuclear instance segmentation. Using BlastoSPIM, we benchmark the performance of five convolutional neural networks and identify Stardist-3D as the most accurate instance segmentation method across preimplantation development. Stardist-3D, trained on BlastoSPIM, performs robustly up to the end of preimplantation development (> 100 nuclei) and enables studies of fate patterning in the late blastocyst. We, then, demonstrate BlastoSPIM's usefulness as pre-train data for related problems. BlastoSPIM and its corresponding Stardist-3D models are available at: blastospim.flatironinstitute.org.

show abstract

Section: Benchmarking Of Seven Instance Segmentation Methods On Blast...mentioning

confidence: 99%

A novel ground truth dataset enables robust 3D nuclear instance segmentation in early mouse embryos

Nunley

Shao

Grover

et al. 2023

Preprint

View full text Add to dashboard Cite

show abstract

“…Whole-well MIPs of the DAPI-channel were segmented using different neural-network-based segmentation workflows based on the fixation timepoint after single cell seeding. Samples fixed and imaged 2 days after seeding of single cells were segmentation with a custom model using RDCNet (version 0.1, 88 ). Samples imaged at all other timepoints after single cell seeding were segmented in CellPose 89,90 (version 2.2, run in Python 3.8.17) using multiple custom-trained models modified from the cyto2 model.…”

Section: Methodsmentioning

confidence: 99%

Dynamics and plasticity of stem cells in the regenerating human colonic epithelium

Oost,

Kahnwald,

Barbiero

et al. 2023

Preprint

Self Cite

View full text Add to dashboard Cite

SummaryThe human intestinal epithelium is a tissue with rapid turnover. Its complex regenerative process and differentiation trajectories have been challenging to study due to its inaccessibility and lack of temporal sampling. To this end, we developed a workflow to culture adult stem cell-derived human intestinal organoids from single cells to maturation. Extensive characterization of our model system indicated a transient regenerative response, followed by differentiation into all mature cell lineages. This switch is accompanied by a transition between two stem cell states. High-content screening and comparison toin vivostudies revealed that an initial fetal-like state is crucial for achieving successful regeneration, while the subsequent adult-like state is vital for maintaining a balance of cell lineages and continuous support of crypt-morphogenesis. Taken together, this study highlights the extensive plasticity of the intestinal epithelium and paves the way for further studies of human intestinal regeneration and its deregulation in pathologies.

show abstract

“…Our main initial motivation was to test whether we could incorporate the spatial information from the lineage trees spots for training segmentation models. To that end, we decided to use the RDCNet instance segmentation network as a base 22 , taking advantage of its inherent recursive architecture. First, nuclei are segmented in 3D following a deep learning model trained with a mix of complete and partial annotations.…”

Section: Dedicated Image Processing Workflowmentioning

confidence: 99%

Multiscale light-sheet organoid imaging framework

et al. 2022

Self Cite

View full text Add to dashboard Cite

Organoids provide an accessible in vitro system to mimic the dynamics of tissue regeneration and development. However, long-term live-imaging of organoids remains challenging. Here we present an experimental and image-processing framework capable of turning long-term light-sheet imaging of intestinal organoids into digital organoids. The framework combines specific imaging optimization combined with data processing via deep learning techniques to segment single organoids, their lumen, cells and nuclei in 3D over long periods of time. By linking lineage trees with corresponding 3D segmentation meshes for each organoid, the extracted information is visualized using a web-based “Digital Organoid Viewer” tool allowing combined understanding of the multivariate and multiscale data. We also show backtracking of cells of interest, providing detailed information about their history within entire organoid contexts. Furthermore, we show cytokinesis failure of regenerative cells and that these cells never reside in the intestinal crypt, hinting at a tissue scale control on cellular fidelity.

show abstract

RDCNet: Instance Segmentation with a Minimalist Recurrent Residual Network

Cited by 7 publications

References 20 publications

A novel ground truth dataset enables robust 3D nuclear instance segmentation in early mouse embryos

A novel ground truth dataset enables robust 3D nuclear instance segmentation in early mouse embryos

Dynamics and plasticity of stem cells in the regenerating human colonic epithelium

Multiscale light-sheet organoid imaging framework

Contact Info

Product

Resources

About