Who will likely develop Alzheimer’s disease? How can it be detected in the brain long before symptoms appear? Dr. Maged Goubran is one of the scientists at Sunnybrook Research Institute (SRI) conducting multiple leading-edge studies that, together, will help answer these questions.
By developing advanced machine learning and artificial intelligence (AI) models and applying them to some of the largest data sets in the world, Dr. Goubran’s team is learning to detect, with greater precision than ever before, cognitive decline years before symptoms arise. Also at the heart of this work is discovering what minute structural and functional changes and network breakdown within the brain can be used as biomarkers, or early/accurate indicators, of neurodegenerative disorders.
“A key missing piece in neurodegeneration research are the algorithms that learn from large amounts of individual data from magnetic resonance imaging (MRI), PET scanning, genomic analysis, and demographics, combined with results from cognitive testing, that can be applied in the clinic,” says Dr. Goubran, a scientist in the physical sciences platform at SRI. “This work is of critical importance because it can provide urgently needed guidance on prevention and personalized treatment decisions.”
Dr. Maged Goubran
Sunnybrook Research Institute
Alzheimer’s disease is one of Canada’s greatest healthcare challenges, currently having a devastating emotional and physical burden (directly or indirectly) on more than 1.1 million people. This number is expected to rise dramatically in the next 20 years. The disease has an estimated health care cost of $10.4 billion annually.
A big focus of Alzheimer’s research and care has been on preventing symptoms earlier in the progression of disease. This tactic relies on identifying healthy adults at high risk of future cognitive decline; these people may have small changes in their brains starting a few decades before the first symptoms appear. While much research has been done in this area, Dr. Goubran and his team are improving on previous approaches in several ways: by building more powerful computational models than in the past, by using larger, more robust population data sets, by focusing on individual rather than group-level predictions (to address the large patient variability) and by pinpointing Alzheimer’s progression in asymptomatic rather than only symptomatic individuals.
Tracking progression of disease
Dr. Goubran is building on earlier research that establishes shrinkage of the hippocampus as a possible early sign of dementia. The hippocampus is the brain’s centre for memory and navigation; it is often one of the first areas to be impacted by Alzheimer’s disease.
The team is developing AI algorithms to map changes in not just the volume but also the shape of the hippocampi (and its subdivisions), as well as other important structures and vascular lesions of the brain during aging. The models are informed by hundreds of brain MRIs from multiple studies including the Sunnybrook Dementia Study led by Dr. Sandra Black, senior scientist and director of the Sandra Black Centre for Brain Resilience and Recovery at SRI. The researchers will use these imaging biomarkers and a normative population of thousands of Canadians to track and predict patient progression.
Others around the world will benefit from these AI techniques, as the team will validate and share them as open-source tools. “We plan to make our algorithms publicly available and easy to use for the research community,” says Dr. Goubran. For patients and clinicians, this could mean improved diagnosis, customized treatments, and better ways to monitor disease-modifying therapies currently being studied.
Mapping network changes
Another area of his research involves studying how well different brain regions are connecting with one another. Altered connectivity could signal problems down the road. “In neurodegenerative diseases like Alzheimer’s, a network disorder, there are a lot of changes in function and structure of brain networks, so we’re doing also a lot of work on that front, trying to develop novel signatures of network dysfunction,” says Dr. Goubran.
This research direction has two foci. One is preclinical work in the laboratory to develop newer signatures (read-outs) of network dysfunction in Alzheimer’s models. The other is developing new computational techniques to analyse functional and diffusion MRI scans in order to better understand and map network changes in-vivo and develop biomarkers. Dr. Goubran was recently awarded a New Investigator Grant from the Alzheimer Society of Canada to help support this work. He collaborates closely on these projects with Dr. Bojana Stefanovic, Dr. JoAnne McLaurin, Dr. Jennifer Rabin, and Dr. Black, as well as groups at McGill and Harvard universities.
The hope is that when altered connectivity between brain regions is detected, a combination drug treatment that targets abnormal buildup of two proteins involved in Alzheimer’s disease – amyloid and Tau – will slow deterioration of key cognitive networks or, if caught early enough in the pre-symptomatic stage, may be able to restore normal connectivity and cognition.
“We’re hoping our work will lead to larger scale efforts to develop combined therapies that will eventually get to the clinic and really push the development of personalized medicine for neurodegenerative diseases,” says Dr. Goubran.