@article{8f8524073e9a43278ce05afa4c3fa2b1,
title = "Interpretable Deep Learning Uncovers Cellular Properties in Label-Free Live Cell Images that are Predictive of Highly Metastatic Melanoma",
abstract = "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 employ a generative neural network in combination with supervised machine learning to classify patient-derived melanoma xenografts as “efficient” or “inefficient” metastatic, validate predictions regarding melanoma cell lines with unknown metastatic efficiency in mouse xenografts, and use the network to generate in silico cell images that amplify the critical predictive cell properties. These exaggerated 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. A record of this paper's transparent peer review process is included in the supplemental information. Video Abstract: [Figure presented]",
keywords = "interpretable deep learning, live cell imaging, melanoma metastasis",
author = "Assaf Zaritsky and Jamieson, {Andrew R.} and Welf, {Erik S.} and Andres Nevarez and Justin Cillay and Ugur Eskiocak and Cantarel, {Brandi L.} and Gaudenz Danuser",
note = "Funding Information: This work was supported by the Cancer Prevention and Research Institute of Texas (CPRIT R160622 to GD), the National Institutes of Health ( R35GM126428 to G.D.; K25CA204526 to E.S.W.), and the Israeli Council for Higher Education (CHE) via the Data Science Research Center, Ben-Gurion University of the Negev , Israel (to A.Z.). We thank Sean Morrison for PDX-derived cell models. We thank Andrew R. Cohen for LEVER. Funding Information: This work was supported by the Cancer Prevention and Research Institute of Texas (CPRIT R160622 to GD), the National Institutes of Health (R35GM126428 to G.D.; K25CA204526 to E.S.W.), and the Israeli Council for Higher Education (CHE) via the Data Science Research Center, Ben-Gurion University of the Negev, Israel (to A.Z.). We thank Sean Morrison for PDX-derived cell models. We thank Andrew R. Cohen for LEVER. Conceptualization, A.Z. E.S.W. and G.D.; methodology, A.Z. A.R.J. and E.S.W.; software, A.R.J.; formal analysis, B.L.C.; investigation, A.Z. A.R.J. A.N. and J.C.; resources, U.E.; writing ? original draft, A.Z.; writing ? review & editing, A.Z. A.R.J. A.N. E.W.S. and G.D.; supervision, G.D.; funding acquisition, G.D. The authors declare no competing interests. One or more of the authors of this paper self-identifies as an underrepresented ethnic minority in science. One or more of the authors of this paper self-identifies as living with a disability. One or more of the authors of this paper received support from a program designed to increase minority representation in science. Publisher Copyright: {\textcopyright} 2021 The Authors",
year = "2021",
month = jul,
day = "21",
doi = "10.1016/j.cels.2021.05.003",
language = "English",
volume = "12",
pages = "733--747",
journal = "Cell Systems",
issn = "2405-4712",
publisher = "Cell Press",
number = "7",
}