Artificial intelligence can identify changes in the brains of people likely to get Alzheimer’s disease almost a decade before doctors can diagnose the disease from symptoms alone. According to a New Scientist report, the technique uses non-invasive MRI scans to identify alterations in how regions of the brain are connected.
Alzheimer’s is a neurodegenerative disease that is the leading cause of dementia for the elderly, eventually leading to loss of memory and cognitive functions. The race is on to diagnose the disease as early as possible. Although there is no cure, drugs in development are likely to work better the earlier they are given. An early diagnosis can also allow people to start making lifestyle changes to help slow the progression of the disease.
In an effort to enable earlier diagnosis, Nicola Amoroso and Marianna La Rocca at the University of Bari in Italy and their colleagues developed a machine-learning algorithm to discern structural changes in the brain caused by Alzheimer’s disease. First, they trained the algorithm using 67 MRI scans, 38 of which were from people who had Alzheimer’s and 29 from healthy controls. The scans came from the Alzheimer’s Disease Neuroimaging Initiative database at the University of Southern California in Los Angeles.
The report says the idea was to teach the algorithm to correctly classify and discriminate between diseased and healthy brains. The researchers divided each brain scan into small regions and analysed the neuronal connectivity between them, without making any assumptions about the ideal size of these regions for diagnosis.
Alzheimer’s disease has been linked to the formation of sticky beta-amyloid plaques and neurofibrillary tau tangles in the brain. “Nowadays, cerebrospinal fluid analyses and brain imaging using radioactive tracers can tell us to what extent the brain is covered with plaques and tangles, and are able to predict relatively accurately who is at high risk of developing Alzheimer’s 10 years later,” says La Rocca. “However, these methods are very invasive, expensive and only available at highly specialised centres.”
In contrast, the new technique can distinguish with similar accuracy between brains that are normal and the brains of people with MCI who will go on to develop Alzheimer’s disease in about a decade – but using a simpler, cheaper and non-invasive technique. More work will be needed to distinguish between people with MCI whose brains go on to age normally, or who might develop other kinds of dementia.
Blood tests that look for biomarkers of Alzheimer’s could be even cheaper and simpler than the new technique, but none are on the market yet. “There are no blood tests for Alzheimer’s disease,” says Goran Šimić at the University of Zagreb in Croatia. “There have been some attempts, but without much success yet.”
The report says Patrick Hof at the Icahn School of Medicine at Mount Sinai in New York is intrigued by the new test. He says that a method that might predict the disease a decade before it is fully expressed would be “incredibly valuable” should preventative therapeutics emerge.
La Rocca says her team now intends to extend the technique to help with the early diagnosis of other neurodegenerative conditions such as Parkinson’s disease. “It’s a method that is very versatile,” she says.
Analysis and quantification of brain structural changes, using Magnetic resonance imaging (MRI), are increasingly used to define novel biomarkers of brain pathologies, such as Alzheimer’s disease (AD). Network-based models of the brain have shown that both local and global topological properties can reveal patterns of disease propagation. On the other hand, intra-subject descriptions cannot exploit the whole information context, accessible through inter-subject comparisons. To address this, we developed a novel approach, which models brain structural connectivity atrophy with a multiplex network and summarizes it within a classification score. On an independent dataset multiplex networks were able to correctly segregate, from normal controls (NC), AD patients and subjects with mild cognitive impairment that will convert to AD (cMCI) with an accuracy of, respectively, 0.86±0.01 and 0.84±0.01. The model also shows that illness effects are maximally detected by parceling the brain in equal volumes of 3000 mm3 (“patches”), without any a priori segmentation based on anatomical features. A direct comparison to standard voxel-based morphometry on the same dataset showed that the multiplex network approach had higher sensitivity. This method is general and can have twofold potential applications: providing a reliable tool for clinical trials and a disease signature of neurodegenerative pathologies.
Nicola Amoroso, Marianna La Rocca, Stefania Bruno, Tommaso Maggipinto, Alfonso Monaco, Roberto Bellotti, Sabina Tangaro