bioRxiv Science⌕ Search
Home/Archive/bioRxiv

bioRxiv · 10.1101/137406

Brain Development: Machine Learning Analysis Of Individual Stem Cells In Live 3D Tissue

Hailstone, M.·Yang, L.·Waithe, D.·Samuels, T. J.·Arava, Y.·Dobrzycki, T.·Parton, R. M.·Davis, I.
View original source ↗Export metadata ↓

Abstract

A major challenge in cell and developmental biology is the automated identification and quantitation of cells in complex multilayered tissues. We developed CytoCensus: an easily deployed implementation of supervised machine learning that extends convenient 2D \"point- and-click\" user training to 3D detection of cells in challenging datasets with ill-defined cell boundaries. In tests on these datasets, CytoCensus outperforms other freely available image analysis software in accuracy and speed of cell detection. We used CytoCensus to count stem cells and their progeny, and to quantify individual cell divisions from time-lapse movies of explanted Drosophila larval brains, comparing wild-type and mutant phenotypes. We further illustrate the general utility and future potential of CytoCensus by analysing the 3D organisation of multiple cell classes in Zebrafish retinal organoids and cell distributions in mouse embryos. CytoCensus opens the possibility of straightforward and robust automated analysis of developmental phenotypes in complex tissues.\n\nSummaryHailstone et al. develop CytoCensus, a \"point-and-click\" supervised machine-learning image analysis software to quantitatively identify defined cell classes and divisions from large multidimensional data sets of complex tissues. They demonstrate its utility in analysing challenging developmental phenotypes in living explanted Drosophila larval brains, mammalian embryos and zebrafish organoids. They further show, in comparative tests, a significant improvement in performance over existing easy-to-use image analysis software.\n\n\n\nO_FIG O_LINKSMALLFIG WIDTH=198 HEIGHT=200 SRC=\"FIGDIR/small/137406v4_ufig1.gif\" ALT=\"Figure 1\">\nView larger version (93K):\norg.highwire.dtl.DTLVardef@68912corg.highwire.dtl.DTLVardef@11300bdorg.highwire.dtl.DTLVardef@95810corg.highwire.dtl.DTLVardef@14b77a6_HPS_FORMAT_FIGEXP M_FIG C_FIG HighlightsO_LICytoCensus: machine learning quantitation of cell types in complex 3D tissues\nC_LIO_LISingle cell analysis of division rates from movies of living Drosophila brains in 3D\nC_LIO_LIDiverse applications in the analysis of developing vertebrate tissues and organoids\nC_LIO_LIOutperforms other image analysis software on challenging, low SNR datasets tested\nC_LI

Explore related subjects

developmental biologyPreprint

Keep this discovery

BibTeXRIS

Hailstone, M., Yang, L., Waithe, D., Samuels, T. J., Arava, Y., Dobrzycki, T., Parton, R. M., Davis, I.. 2017-05-14. Brain Development: Machine Learning Analysis Of Individual Stem Cells In Live 3D Tissue. https://doi.org/10.1101/137406

Cite the original work for its findings. Save a collection to share your selection of sources.

KEEP EXPLORING

Related preprints

bioRxiv2026

Neogenin-1 marks myeloid-primed fetal hematopoietic stem cells that undergo progressive lineage-restriction with age

During aging, hematopoietic stem cells (HSCs) increasingly shift from balanced to myeloid-biased differentiation, resulting in reduced lymphoid output and impaired adaptive immunity. The question of whether this lineage bias is established in a subset of HSCs during early development or primarily emerges with aging warrants further investigation. Here, we investigate whether myeloid-biased HSCs (my-HSCs) are established at the fetal liver stage by specifically examining Neogenin-1 (NEO1), a previously defined marker of my-HSCs. We identify two distinct populations of Hoxb5+ HSCs in the fetal liver: NEO1+ and NEO1-, with NEO1+ HSCs exhibiting transcriptional and functional characteristics consistent with my-HSCs. With age, my-HSC-associated transcriptional programs become increasingly reinforced across the Hoxb5+ pHSC compartment, with NEO1+ cells showing early enrichment of this program and both NEO1+ and NEO1- cells acquiring broader myeloid-biased features in aging. These findings suggest that lineage programming can begin early in development and is further shaped by age-related changes, potentially contributing to the functional decline observed in the aging hematopoietic system.

developmental biology↗
bioRxiv2026

Distinct roles for partially redundant transcription factors in Caenorhabditis elegans mesoderm lineage development

Developmental transcription factors often have overlapping functions, making it difficult to define the distinct roles of individual factors during lineage specification. We investigated the partially redundant transcription factors TBX-35 and CEH-51 in the Caenorhabditis elegans embryonic MS mesodermal lineage using 4D lineage tracing, reporter imaging, genetics, and single-cell RNA sequencing. In tbx-35 mutants, MS descendants showed progressively slower cell cycles and a division pattern that increasingly resembled the cousin C lineage. Fate-regulator expression also shifted toward C-like features, including ectopic pal-1 and expanded HLH-1 expression, although mutant cells did not simply adopt normal C-lineage positions. Loss of tbx-35 also impaired a later MS-dependent Notch induction in the AB lineage while leaving an earlier induction intact. CEH-51 showed a different pattern of activity whereby its protein became enriched in anterior MS daughters, and ceh-51 mutants produced later, more restricted lineage defects that were strongest in descendants of cells with higher CEH-51 levels. Single-cell profiling identified overlapping but nonidentical sets of genes dependent on the two factors. TBX-35-dependent changes were strongest at earlier stages, whereas CEH-51-dependent genes became more prominent later and were enriched in anterior MS sublineages. Finally, temperature-shift experiments determined that the severity and onset of tbx-35 mutant phenotypes depend on the maternal temperature environment and cannot be explained by differences in residual CEH-51 expression. These findings reveal that TBX-35 and CEH-51 contribute differently across the MS lineage and that reliable mesoderm development is supported by overlapping zygotic and maternal regulatory inputs.

developmental biology↗
bioRxiv2026

Dynamic microtubules drive yolk-cytoplasm segregation in the syncytial Drosophila embryo

Yolk-cytoplasm segregation is among the earliest spatial organization events in the developing embryo of many oviparous animals. The segregation process is intimately linked to early embryonic cleavage and pattern formation, and exhibits a wide range of spatial and temporal diversity. However, the underlying cytoskeletal mechanism remains largely unknown, except for a small number of species. Using quantitative live imaging, we investigated yolk segregation in the Drosophila embryo during the syncytial nuclear cycles 11-14. We find that the yolk vesicles move progressively inward in spatial and temporal coordination with the inward expanding microtubule networks that are nucleated from centrosomes positioned at the cortex, whereas cortical actin meshwork remains spatially restricted. Using the gnu RNAi embryo to decouple nuclear migration and division from cytoskeletal dynamics, we establish causality with targeted pharmacological disruption and find that microtubule dynamics is required for yolk segregation, while depolymerization of actin has no discernible effect. In support of a mechanism of growth-propelled passive displacement, microtubule plus end comets come in apparent contact with yolk vesicles, and injected, inert microbeads are displaced towards the embryo center presumably by the same pushing force. These findings identify microtubule polymerization as a predominant driver of yolk-cytoplasm segregation in Drosophila and suggest that diverse cytoskeletal mechanisms evolved to accomplish this crucial reorganization process

developmental biology↗