Alzheimer’s disease (AD) is a multifactorial disease and therefore, several tests are usually performed to evaluate and measure all the critical factors that contribute to its progression. The three most known features that characterise AD are the accumulation of amyloid and tau proteins, and the structural brain changes related to neurodegeneration or brain atrophy. These are related to the brain’s structure and function and can be identified by using brain scans. However, the diagnostic accuracy of brain scans increases when combined with other tests, such as blood, memory or cognitive tests. In a new study published in the Nature Portfolio Journal, Communications Medicine, a team of researchers led by Eric O. Aboagye of the Imperial College London (London, UK) presents a new approach based on a single brain scan able to predict whether a person has AD. This new approach uses magnetic resonance imaging (MRI) brain scans in combination with an artificial intelligence algorithm. To develop this novel approach, the researchers proceeded to use brain scans and segment the brain images into 115 regions. Subsequently, they extracted more than 650 structural features for each region related to size, shape, intensity and texture, among others. To identify the changes to these features and determine whether these changes could predict AD, the researchers first trained the algorithm on brain scans obtained from the Alzheimer’s Disease Neuroimaging Initiative (ADNI).
They then tested the algorithm on brain scans from four different cohorts which were divided into two groups. The first one was the control group and included brain scans from people with diseases unrelated to AD (Frontotemporal Dementia and Parkinson’s disease), and healthy controls. The second group, named the disease group, consisted of people with AD-related Mild Cognitive Impairment (MCI) and AD. Overall, this unsupervised and MRI-based algorithm accurately predicted an AD-related pathology in 98% of the cases and distinguished between early and late stages of AD in 79% of them. Furthermore, the researchers identified changes in areas of the brain not previously associated with AD. This new approach suggests that the MRI-based algorithms may improve the information that clinicians usually obtain from brain scans, by identifying individuals at the early stages of the disease and brain areas possibly involved in or affected by the progression of the AD. https://www.nature.com/articles/s43856-022-00133-4
