ABOVE: Synthetic melanoma samples generated by the neural network
ASSAF ZARITSKY, GAUDENZ DANUSER, ANDREW JAMIESON, ERIK WELF, ANDRES NEVAREZ

Researchers are presenting new images from their deep neural network that analyzes melanoma cells at this year’s American Society for Cell Biology / EMBO meeting, which started Saturday (December 7).

Assaf Zaritsky, now at Ben-Gurion University of the Negev, began this work as a postdoc in Gaudenz Danuser’s lab at the University of Texas Southwestern Medical Center. He and Danuser, along with colleagues Erik Welf and Andrew Jamieson at UT Southwestern and Andres Nevarez at University of California San Diego, developed a deep neural network that uses machine learning to distinguish melanoma cells that have high versus low metastatic potential—meaning how likely the cancer is to spread. This live-cell histology relies on subtle cues from cell actions rather than the traditional method of classifying the cells by shape...

Image Container (remove this text once done styling the container)
Insert Image Here (remove this text once you've added an image)
Cells generated by the neural network morph between different levels of metastatic efficiency.
ASSAF ZARITSKY, GAUDENZ DANUSER, ANDREW JAMIESON, ERIK WELF, ANDRES NEVAREZ
Image Container (remove this text once done styling the container)
Insert Image Here (remove this text once you've added an image)
Cells transition from low to high metastatic efficiency.
ASSAF ZARITSKY, GAUDENZ DANUSER, ANDREW JAMIESON, ERIK WELF, ANDRES NEVAREZ

See “Image of the Day: Bad Behavior”

Emily Makowski is an intern at The Scientist. Email her at emakowski@the-scientist.com.

Interested in reading more?

The Scientist ARCHIVES

Become a Member of

Receive full access to more than 35 years of archives, as well as TS Digest, digital editions of The Scientist, feature stories, and much more!
Already a member?