The study stemmed from work conducted by the group's pathology colleagues in garnering the genotype data of several patients' gastrointestinal cancers, said presenter Dr. Siddharth Govindan. The researchers obtained the radiology reports of these patients and applied natural language processing (NLP) analysis to determine if there were statistical differences in the frequency of certain words in these reports for patients with mutated forms of cancer, compared with those who had wild-type colon cancer.
"As healthcare is moving into the arena of personalized medicine, understanding how different genotypes dictate the course of an otherwise generic cancer could be useful in treating a specific patient," Govindan told AuntMinnie.com. "Today, digital medical records are becoming so voluminous that the information within them often goes untapped, and sensibly using technologies to parse through the data could be useful in many circumstances, including the surfacing of otherwise buried insights."
The study results suggest that different colon cancer genotypes have different imaging findings, potentially leading to different clinical courses, Govindan said.
"More generally, it speaks to a potential utility of NLP in population analytics to understand disease entities," he said.