Format

Send to

Choose Destination
Cell. 2018 Apr 19;173(3):792-803.e19. doi: 10.1016/j.cell.2018.03.040. Epub 2018 Apr 12.

In Silico Labeling: Predicting Fluorescent Labels in Unlabeled Images.

Author information

1
Google, Inc., Mountain View, CA 94043, USA. Electronic address: ericmc@google.com.
2
Google, Inc., Mountain View, CA 94043, USA.
3
Taube/Koret Center for Neurodegenerative Disease Research and DaedalusBio, Gladstone Institutes, San Francisco, CA 94158, USA.
4
Department of Stem Cell and Regenerative Biology, Harvard University, Cambridge, MA 02138, USA; Department of Biomedical Informatics, Harvard Medical School, Boston, MA 02115, USA; Center for Assessment Technology and Continuous Health, Massachusetts General Hospital, Boston, MA 02114, USA.
5
Department of Stem Cell and Regenerative Biology, Harvard University, Cambridge, MA 02138, USA.
6
Google, Inc., Mountain View, CA 94043, USA; Montreal Institute of Learning Algorithms, University of Montreal, Montreal, QC, Canada.
7
Google, Inc., Mountain View, CA 94043, USA; Department of Electrical Engineering, Stanford University, Stanford, CA 94305, USA.
8
Google, Inc., Mountain View, CA 94043, USA. Electronic address: pqnelson@google.com.
9
Taube/Koret Center for Neurodegenerative Disease Research and DaedalusBio, Gladstone Institutes, San Francisco, CA 94158, USA; Departments of Neurology and Physiology, University of California, San Francisco, 94158, USA. Electronic address: sfinkbeiner@gladstone.ucsf.edu.

Abstract

Microscopy is a central method in life sciences. Many popular methods, such as antibody labeling, are used to add physical fluorescent labels to specific cellular constituents. However, these approaches have significant drawbacks, including inconsistency; limitations in the number of simultaneous labels because of spectral overlap; and necessary perturbations of the experiment, such as fixing the cells, to generate the measurement. Here, we show that a computational machine-learning approach, which we call "in silico labeling" (ISL), reliably predicts some fluorescent labels from transmitted-light images of unlabeled fixed or live biological samples. ISL predicts a range of labels, such as those for nuclei, cell type (e.g., neural), and cell state (e.g., cell death). Because prediction happens in silico, the method is consistent, is not limited by spectral overlap, and does not disturb the experiment. ISL generates biological measurements that would otherwise be problematic or impossible to acquire.

KEYWORDS:

cancer; computer vision; deep learning; machine learning; microscopy; neuroscience; stem cells

PMID:
29656897
PMCID:
PMC6309178
DOI:
10.1016/j.cell.2018.03.040
[Indexed for MEDLINE]
Free PMC Article

Supplemental Content

Full text links

Icon for Elsevier Science Icon for PubMed Central
Loading ...
Support Center