Glioma, a form of brain tumor, requires both neurosurgeons and pathologists to work together to provide a proper prognosis to a patient, on how long the treatment will go on.  Although, genomic tests are accurate to an extent, the microscopic work done by pathologists is subjective in nature, often resulting in difference in opinions for the same tumor case among various pathologists (doctors who undergo years of training in this specialised field).

A research project led by a neuropathologist (Daniel J. Brat, MD, PhD), with collaboration with Emory University and the Winship Cancer Institute, have developed an AI system which can predict the course of a patient’s disease more accurately than the doctors.

This innovative approach uses deep learning – a branch of AI – which when adequately trained – can classify images (microscopic images in this case)  accurately when compared to trained human eyes.  Add curnching genomic data to this equation, diagnosis becomes more accurate.  Its been long since we have been using AI in various forms in commercial applications (google’s web searches, or amazon’s recommendations etc), but one the best yield can be achieved when applying this to health care.  Remember that AI is still not a GPT (General Purpose Tech), yet.  Internet was not a GPT until it become and got embraced globally.  Until AI becomes GPT, fears surrounding the use of it more generally will continue to permeate.

Read more:
Source: Machines See the Future for Patients Diagnosed With Brain Tumors

Image courtesy – rdmag.com and Emory Health Sciences

blogpost from girishnair.com