10.17867/10000161
Zaritsky, Assaf
Assaf
Zaritsky
https://orcid.org/0000-0002-1477-5478
Homologous chromosomes undergo sequential quasi-stable interactions during meiotic prophase
University of Dundee
2021
Image
FOS: Biological sciences
Jamieson, Andrew
Andrew
Jamieson
https://orcid.org/0000-0001-5416-9379
Welf, Erik
Erik
Welf
https://orcid.org/0000-0002-5968-3323
Nevarez, Andres
Andres
Nevarez
https://orcid.org/0000-0002-8337-6454
Cillay, Justin
Justin
Cillay
Eskiocak, Ugur
Ugur
Eskiocak
Cantarel, Brandi
Brandi
Cantarel
https://orcid.org/0000-0003-4126-4478
Danuser, Gaudenz
Gaudenz
Danuser
https://orcid.org/0000-0001-8583-2014
2021-05-25
en
https://github.com/IDR/idr0109-zaritsky-melanoma/blob/HEAD/experimentA/idr0109-experimentA-annotation.csv
10.1016/j.cels.2021.05.003
nd2
Creative Commons Attribution 4.0 International
Deep learning has emerged as the technique of choice for identifying hidden patterns in cell imaging data, but is often criticized as ‘black-box’. Here, we demonstrate that a generative neural network captures subtle details of cell appearance, which permit the prediction of the metastatic efficiency of patient-derived melanoma xenografts with known clinical outcomes. To probe the predictor, we used the network to generate “in-silico” cell images that amplified the critical predictive cellular features. These images unveiled pseudopodial extensions and increased light scattering as hallmark properties of metastatic cells. We validated this interpretation using live cells spontaneously transitioning between states indicative of low and high metastatic efficiency. This study illustrates how the application of Artificial Intelligence can support the identification of cellular properties that are predictive of complex phenotypes and integrated cell functions, but are too subtle to be identified in the raw imagery by a human expert.
Live melanoma single cell phase contrast imaging with multiple labels: cell category (PDX, cell line, melanocyte, clonal expansion), PDX metastatic efficiency (high, low, unknown), cell type (e.g. A375, MV3, m405).
Cancer Prevention and Research Institute of Texas
https://doi.org/10.13039/100004917
CPRIT R160622
National Institutes of Health
https://doi.org/10.13039/100000002
R01GM071868
The Mechanics of Actin-mediated Cell Protrusion
National Institutes of Health
https://doi.org/10.13039/100000002
K25CA204526