Researchers say the technology focuses on what is known as “spatial” gene expression, which examines how genetic activity can vary across different areas of the same tumor.
Researchers at Cedars-Sinai say they have developed an artificial intelligence tool that could reduce the time and cost required to analyze the genetic activity of cancer tumors, a step scientists say may eventually help expand access to personalized cancer treatment.
The tool, called Path2Space, was described this week in the scientific journal Cell. According to the study, the system uses AI to predict gene expression patterns within tumors based on digital images from biopsy slides.
Researchers say the technology focuses on what is known as “spatial” gene expression, which examines how genetic activity can vary across different areas of the same tumor. Scientists believe understanding those variations may help explain why some cancers respond differently to treatment.
Current spatial gene expression profiling methods can take weeks and cost thousands of dollars, according to the study authors. The new AI-based method produces results in minutes using standard pathology images.
The research team trained the model using breast cancer biopsy data and later tested it on additional patient datasets to evaluate its accuracy. Investigators said the system was able to predict the activity of thousands of genes across tumor samples with results that closely matched laboratory measurements.
Lead researcher Eytan Ruppin said the technology could help scientists study larger numbers of tumor samples than traditional methods currently allow.
Researchers also said the tool may help identify biomarkers tied to treatment response or poor outcomes, though they emphasized the system still requires further validation through clinical trials before it could be used in routine patient care.
