AI and neuroscience

How BrainSightAI combines AI and neuroscience in psychiatric investigation

BrainSightAI, with the help of artificial intelligence, provides faster and deeper insights to help earlier detection of psychotic disorders.

Much has been spoken about mental health in the recent years. Courtesy technological advancements there is now adequate research, investment and deployment in diagnosing and treating mental health issues. One such startup contributing to the space is BrainSightAI. Founded by Laina Emmanuel and Dr. Rimjhim Agrawal, BrainSightAI combines artificial intelligence and neuroscience to enable greater precision in neurological and psychiatric investigation for accelerated patient outcomes.

AI’s role in the diagnosis process:

Laina Emmanuel, CEO, BrainSightAI, explains a psychotic disorder should be detected within three months. “There is a golden critical period in which you require the patient to be on the correct medicine, otherwise the probabilities of relapse, etc. are much higher,” she says.

Rimjhim Agrawal, co-founder and CTO of BrainSightAI, explains the process BrainSightAI employs to diagnose ailments. “We have a complex dataset called the resting state MRI, which is a video of the brain activity for 7 minutes,” he says.

BrainSightAI processes a 4D dataset. The primary goal is eliminating the noise, get maximum signals and that dataset gets ingested in the machine algorithm. What comes out is the connection and the activity of the brain, defining the changes in the connectivity and activity of the person having a certain disorder.

“For instance, the controls will have higher activity in one part of the brain but it will be suppressed because of the disorder. That is what we give in detail to the doctor. It is not exactly the diagnosis; it is the features that we define that is important for the doctor to diagnose a certain disorder and make more informed decision,” says Rimjhim.

The system is not without its challenges, but Laina and Rimjhim have found a way to simplify it. One major challenge for resting state MRI is that it is highly complex where multiple steps and stages need clean data. But the processing itself takes a long time and as a researcher in any lab, mostly resting state is used in research setting where there are PhDs defining the pipeline for a couple of years.

Once the pipeline has been standardized, then the signal is extracted. Rimjhim and Laina are aware of how tedious it is to get a standardized pipeline as it took them a couple of years to get it right with the golden standards of processing. Later, at BrainSightAI, they automated everything so no doctor will need a PhD or researcher to specifically sit on that and they can start their work from more advanced level of doing the analysis, so they don’t need to waste their one or two years in standardization.

Another challenge solved by BrainSightAI was using the signal for classification and currently mapping several brain activities for surgical planning and analysis of certain disorders, finding out individualized pattern, etc.

Classification of disorders include schizophrenia versus bipolar disorder. Both are psychosis disorders with symptoms of hallucinations, delusions, paranoia so it becomes difficult at times to diagnose in these certain categories. The patient might show the same symptoms of psychosis and might have different origin. Hence, giving details whether a person belongs to a schizophrenia category or whether a person has more features of bipolar disorder and if they require mood stabilizer along with the anti-psychotics, these are the decisions that the doctors can make.

“Another use case is to map the brain activity for minimal functioning loss in pre surgical planning. It is finding out the grey area with the presence of tumor and how those brain area functions are going to be impacted because of the tumor. And if there is a surgery what would be the better path or area to approach the tumor. Those are the things we are working closely with the neurosurgeons and giving them the reports,” says Laina.

Benefits of the solution:

Laina uses three words to describe the benefits derived: Faster, deeper, richer. “The solution provides deeper insight into the patient’s response and activity with the pre surgical planning and is much more insightful and early as well. For instance, in dementia, all the other techniques like eye scan or virtual reality or people working on other tech, it usually comes at around 3-4 years or 5-6 years after changes in your brain have started happening. With the functional connectivity and analysis, you are actually able to catch much earlier on year 0,1,2. Therefore, it is earlier and faster,” she says.

Roadblocks for acceptance:

The solution could potentially help diagnose the psychotic disorders in a much faster manner than the present prevalent methodology. But what’s stopping it from widespread adoption. Laina states resting state FMRI is not used in every major research or hospital. The reason being the difficulty to process.

“A lot of doctors across the world, despite the potential, don’t use it and that’s what we want to solve with our product, which is a one click solution. Doctors just have to upload the FMRI and get the result within 40 minutes to one and half hour, depending on the complexity of the data and processing wanted, as easy as any other imaging technology,” says Laina.

Deep tech ecosystem in the country:

Laina believes the risk appetite for deep tech has been low. “The ecosystem around something as deep tech as this is not as developed as for consumer goods or apps. It surprises me because this technology has so much power than a simple app and why aren’t there enough people thinking about how you bring this into practice and enough people putting bets on something like this,” she says.

Rimjhim states in the deep tech ecosystem, the processes are also slower. The government t is trying to provide good space for deep tech community. However, the overall process is still slow. If the govt efforts are accelerated or there is more space to have connection it can expedite the process,” she says.

Both Laina and Rimjhim agree that the accelerator ecosystem helped with the kind of computational resources that they already have which is difficult for a startup to get access to. “It would be useful if corporate hospitals have translational research units which is quite common for hospitals outside India, where they are working with startups to develop the data and see what hypothesis can be tested through AI,” Laina concludes.

Leave a Reply

Your email address will not be published. Required fields are marked *