IT researchers at MIT are developing a computer system that will be able to predict the effects of brain disease on the anatomy of the brain. By using genetic, demographic and clinical data from patients with neurodegenerative diseases, the computer has been programmed to predict the development of brain diseases and disorder with more accuracy than just using data from MRI scans alone.
The research, led by Polina Golland, a professor of electrical engineering and computer science, used data from the Alzheimer’s Disease Neuroimaging Initiative, that included MRI scans of the same patients taken at various stages of the disease. Each three-dimensional scan was made up of millions of tiny cubes or voxels; these scans were then combined to produce a generic brain template. Two experiments were conducted: the first experiment involved training their computer system on scans of both healthy patients and those displaying some evidence of mild cognitive impairment or Alzheimer’s disease. The second experiment trained their computer system with data obtained only from healthy subjects.
In the first experiment the system was trained twice in the second experiment the system was trained once
During the first experiment, the system was trained twice. The first training session used just the MRI scans, while the second training session was supplemented with additional information including genetic markers, demographic data and some basic clinical data. In the second experiment, the system was trained just once, using both the MRI data and the supplementary data of healthy subjects. However, in this experiment, they used it to predict what they brains of Alzheimer’s patients would look like if they had not been affected by the disease. The researchers believe that while there is no clinical data to validate these predictions, this kind of exploration could be useful as it shows how changes in individual patients with brain disease evolve, compared to the progression exhibited by normal degeneration.
This work could prove extremely useful in the future, as ultimately it could avoid the need for patients to undergo MRI and PET scans when evaluating the potential of experimental Alzheimer’s drugs. Both MRI and PET scans are extremely expensive, so if a computer system can be developed to replace these systems when evaluating patients, it could prove to be very beneficial.
Neuro News is an online social site dedicated to providing you with all the important information you need regarding a variety of different neurological conditions. We have the goal of creating a growing internet platform that anyone can use to connect with others and share ideas, knowledge and resources relating to neurological conditions.
Looking for a job within the Neurological conditions care industry? See our Careers page for exclusive job opportunities at some of the best care facilities.