Research

Cognitive Neuroscience of Medicine

Physicians make hunderds of decisions every day, many of which are made under extreme pressure. Their brains are constantly processing new information, integrating it with everything they have learned, and using it to produce conscious decisions that lead to the best possible medical outcome. Given the immense workload that physicians undertake, as well as the high level of difficulty of the tasks they perform, it is not surprising that errors in medicine are quite prevalent. These errors, however, may be due to the overreliance on physicians' strained and overworked conscious systems. What if, instead, their non-conscious systems could tell us more?

We are interested in how physicians' non-conscious brain systems process information, how they encode medical experience, and how they might guide the perceptual system towards the truth. We intend to use this understanding across multiple medical tasks to extract and reveal non-conscious signals, in hopes that they will steer us, as well as the physician to whom they belong, towards the best possible medical decision. Our goal is to aid physicians' conscious report by showing them their own non-conscious report.

Currently, our major area of focus is in diagnostic radiology, specifically in harnessing the perceptual processing power of radiologists' non-conscious systems to detect missed lung nodules in chest CTs. We used an eyetracking camera to record radiologists' eye position while they read chest CTs. We were able to show that, even when radiologists reported not seeing a lung nodule (the location of which was know to us), their eyes dwelled on the nodules for more time than normal tissue, their pupils were larger when looking at nodules than when looking at normal tissue, and they looked at nodules more times than anywhere else on the image. With our collaborators, we fed the radiologists' eyetracking data through a Machine Learning, which was then able to identify the locations of additional lung nodules, which the radiologist had previosuly consciously "missed". We hope to utilize this technique to aid radiologists in their diagnostic abilities, and to apply it across multiple other medical domains.

This research is supported by an R01 from the National Cancer Institute at the National Institutes of Health.