Mar 16 2021
In the case of Alzheimer’s disease (AD), warning signs can start in the brain region years before the appearance of the first symptoms.
Image Credit: Peshkova/shutterstock.com
Detecting such signs may enable lifestyle changes that could probably prolong the damage caused to the brain by Alzheimer’s disease.
Improving the diagnostic accuracy of Alzheimer's disease is an important clinical goal. If we are able to increase the diagnostic accuracy of the models in ways that can leverage existing data such as MRI scans, then that can be hugely beneficial.
Vijaya B. Kolachalama, PhD, Study Corresponding Author and Assistant Professor of Medicine, Boston University School of Medicine
Kolachalama and his research team used a sophisticated artificial intelligence (AI) framework based on a game theory (called generative adversarial network or GAN) and successfully processed the images of the brain (some images had high and low quality) to create a model that is capable of classifying Alzheimer’s disease with enhanced precision.
The quality of an MRI scan is based on the type of scanner instrument used. For instance, a 1.5 Tesla magnet scanner has a somewhat lower quality image when compared to an image captured from a 3 Tesla magnet scanner.
Magnetic strength is a major parameter correlated to a particular scanner. The team acquired MR images of the brain from both the 3 Tesla and the 1.5 Tesla scanners of the same subjects taken simultaneously and designed a GAN model that learned from both sets of images.
As the GAN model was 'learning' from the images taken by the 3 Tesla and 1.5 Tesla scanners, it produced images that had better quality when compared to the 1.5 Tesla scanners.
These generated images were also better at estimating the status of Alzheimer’s disease on these patients than what could probably be achieved by using the models based only on 1.5 Tesla images.
Our model essentially can take 1.5 Tesla scanner derived images and generate images that are of better quality and we can also use the derived images to better predict Alzheimer's disease than what we could possibly do using just 1.5 Tesla-based images alone.
Vijaya B. Kolachalama, PhD, Study Corresponding Author and Assistant Professor of Medicine, Boston University School of Medicine
Throughout the world, the population of people aged 65 and above is growing faster when compared to all other age groups. And by 2050, one in six people worldwide will be over age 65. Although the total estimated healthcare expenses for AD treatment in 2020 were projected at $305 billion, it is anticipated to increase to over $1 trillion as the population ages.
The significant burden on patients and their caregivers, specifically, family caregivers of AD patients experience extreme distress and hardship that represent a significant but usually hidden burden.
The researchers believe that it might be possible to create images of improved quality on disease cohorts that have earlier utilized the 1.5T scanners, and in those centers that still continue to depend on 1.5T scanners.
This would allow us to reconstruct the earliest phases of AD, and build a more accurate model of predicting Alzheimer's disease status than would otherwise be possible using data from 1.5T scanners alone.
Vijaya B. Kolachalama, PhD, Study Corresponding Author and Assistant Professor of Medicine, Boston University School of Medicine
Kolachalama believes that these latest AI techniques can be used optimally so that the medical imaging community could get the best out of the developments in AI technology.
Kolachalama also hopes that these frameworks can be utilized to harmonize imaging data across numerous studies so that models can be designed and compared across various populations. This can result in the development of more improved methods for diagnosing AD.
The study was partly funded by the Karen Toffler Charitable Trust, National Center for Advancing Translational Sciences, National Institutes of Health (NIH), via BU-CTSI Grant (1UL1TR001430), a Scientist Development Grant (17SDG33670323), and a Strategically Focused Research Network (SFRN) Center Grant (20SFRN35460031) from the American Heart Association, and a Hariri Research Award from the Hariri Institute for Computing and Computational Science & Engineering at Boston University, Framingham Heart Study's National Heart, Lung and Blood Institute contract (N01-HC-25195; HHSN268201500001I) and NIH grants (R01-AG062109, R21-CA253498, R01-AG008122, R01-AG016495, R01AG033040, R01-AG054156, and R01-AG049810).
Further support was offered by Boston University Alzheimer's Disease Center (P30-AG013846) and Boston University’s Affinity Research Collaboratives program.
Journal Reference:
Zhou, X., et al. (2021) Enhancing magnetic resonance imaging-driven Alzheimer’s disease classification performance using generative adversarial learning. Alzheimer's Research & Therapy. doi.org/10.1186/s13195-021-00797-5.
Source: https://www.bumc.bu.edu/