Artificial intelligence can dethrone MRI.


With the future enhancements on AI technologies in healthcare, today’s MRI scan might be part of the medical history. AI would be able to determine the possible complexities in the early stages. Marek Kubicki M.D. is a Professor of Psychiatry and Radiology, He is the Associate Director of Psychiatry Neuroimaging Laboratory in Harvard Medical School and he answered my questions.

With the help of artificial intelligence will brain imaging be cheaper and more accessible to the patient?

There are already attempts and projects to make imaging cheaper and more accessible. In the Southern United States, they put imaging machines in the big busses. They go to poorer communities and they do very cheap imaging.
Another way to make imaging more accessible is to reduce the size of the machine.  This way, by reducing the size and making it more compact, transportation would be less of a hustle hence making it more accessible.  

Another way to make imaging more accessible and cheaper would be by improving the performance of the MRI machine which would then reduce the time it takes to do the imaging (it takes now half an hour or more to scan one patient).  Since the single MRI machine costs about a million dollars, faster imaging could serve more patients and reduce the cost per scan.

Alternatively, there are other portable scanning machines that can also help diagnose less complicated matters.

With the help of artificial intelligence, we may not even need MRI machines in the future because the AI would be able to predict or foresee possible complexities beforehand.

Artificial intelligence can dethrone MRI.
Artificial intelligence can dethrone MRI.

How would collected neuro-imaging data or the methodology used to collect this data affect the future of the overall psychiatric diagnostics?

The “imaging community” is working on developing biomarkers of certain diseases.  For example, you put a patient in the scanner and you measure few parameters in these images and they show you that there might be inflammation going on in the brain and the doctor needs to give this particular medication to heal this inflammation. So, that measure would be considered a biomarker.

The biomarkers indicating or pinpointing a signal in the brain could easily be used to predict future complications.

We are investigating the role of neuro-inflammation now in schizophrenia and we think that we have a couple of biomarkers in place. So you can put a subject who is hallucinating in the magnet and we can measure this biomarker that there is or there isn’t inflammation and if there is inflammation all we need to do is to give the anti-inflammatory drugs to the patient and it will heal these hallucinations. And if there is no neuro-inflammation then you go with antipsychotic drugs.

Is the future of these machines advancing in a way to solve or explore the secret parts of the brain in the direction of even with the possibility of creating a new brain?”

There is already a project called “brain-initiative”. Obama helped to initiate that project and it is now funded with millions of dollars, and it is to try to understand better the human brain.

Studying the brain is more difficult than studying any other organ, as the brain can not be kept alive outside of the human body.  This alone is an indication of much work ahead.

Also, a lot of the things that the brain is doing still cannot be measured fully because of the complexity of the brain function.

What would you say that the most exciting latest development in the imaging community?

A group of people located in California and Boston working collaboratively to use nano-particles as a contrast to image proteins in the brain.  However, these tests cannot be conducted on humans yet because these particles are highly toxic.

Once we can do this safely, we could find out much more about the brain dysfunction in psychiatric or neurological diseases.