AI Used to Distinguish Metastatic Potential in Skin Cancers

Researchers from UT Southwestern Medical Center have investigated a new way to use AI to determine the metastatic potential of melanoma.

Researchers from the University of Texas (UT) Southwestern Medical Center developed a process to discover which skin cancers have higher metastatic potential by using artificial intelligence (AI).1 The data demonstrated the opportunity for AI-based tools to change pathology for cancer and other diseases. 

“We now have a general framework that allows us to take tissue samples and predict mechanisms inside cells that drive disease, mechanisms that are currently inaccessible in any other way,” said study leader Gaudenz Danuser, PhD, professor and chair of the Lyda Hill Department of Bioinformatics at UT Southwestern, Dallas, Texas.

Researchers investigated how to use AI technology’s deep learning-based methods to look for differences in disease characteristics that may offer insight on prognoses or guide treatments. The differences distinguished by AI are generally not interpretable in terms of specific cellular characteristics – a drawback that has made AI a tough sell for clinical use according to the source material.1 

So, to overcome this difficulty, the scientists had the AI search for differences between pictures of melanoma cells with high and low metastatic potential and after reverse-engineered to discover the distinguishing characteristics of the images. The images were taken from tumors from 7 patients, of which the researchers knew their past disease history, by taking videos of 12,000 random cells living in petri dishes which generated about 1,700,000 raw images. In total, the AI algorithm found 56 abstract numerical features from this process.

One feature found was able to accurately discriminate between cells with high and low metastatic potential. By manipulating this feature, the researchers produced artificial images that exaggerated visible characteristics inherent to metastasis that human eyes cannot detect according to Danuser. The highly metastatic cells produced slightly more pseudopodial extensions – a type of fingerlike projection – and had increased light scattering, an effect that may be due to subtle rearrangements of cellular organelles the source said.

The researchers first classified the metastatic potential of cells from human melanomas that had been frozen and cultured in petri dishes for 30 years, and then implanted them into mice. Those predicted to be highly metastatic formed tumors that readily spread throughout the animals, while those predicted to have low metastatic potential spread little or not at all.

More trials and studies will be needed before this can become a part of the clinical care setting.

Disclosure:

This study was funded by grants from the Cancer Prevention and Research Institute of Texas (CPRIT R160622), the National Institutes of Health (R35GM126428, K25CA204526), and the Israeli Council for Higher Education via the Data Science Research Center, Ben-Gurion University of the Negev, Israel.

Reference:

1. Artificial intelligence algorithm developed to assess metastatic potential in skin cancers. Accessed August 6, 2021. https://www.utsouthwestern.edu/newsroom/articles/year-2021/ai-melanoma.html