Machine learning can distinguish differences in cancer cells

When physicians are able to discover specific types of cells that contribute to malignant and drug-resistant cancers, patients have a better chance of receiving life-saving care. Researchers from Brown University have developed an image analysis technique that can identify two cell types known to cause tumor progression.

Through a process known as epithelial-mesenchymal transition (EMT), epithelial cells transform into mesenchymal cells. The higher the number of mesenchymal cells within a tumor, the more malignant and resistance to drug therapies it will be. The new technique uses microscopic imaging and machine learning to categorize the two types of epithelial cells into the safe and malignant. The findings were published in the journal Integrative Biology.

"We know that there are these different cell types interacting within tumors and that therapeutics can target these cells differently," said Susan Leggett, a doctoral student in Brown's pathobiology graduate program. "We've developed a model that can pick out these cell types automatically and in an unbiased way. We think this could help us better understand how these different cell types respond to drug treatment."

While the cells can be differentiated by shape, the differences may be subtle enough to escape detection from the human eye. The algorithm is able to pick out the smallest differences at a rate of 92 percent accuracy.

"When we do initial lab testing of drugs, we put cells on a plate, apply the drug and see what lives and what dies," said Ian Y. Wong, assistant professor of engineering at Brown University in Providence, R.I. "This could provide us with a more nuanced picture of the drug's effects and help us to see whether sub-lethal doses may prime cells for resistance."