July 6, 2021 | A team of researchers from Beijing Normal University in China has developed an artificial intelligence-based brain age prediction model to quantify deviations from a healthy brain-aging trajectory in patients with mild cognitive impairment. The work was published in Radiology: Artificial Intelligence (DOI: 10.1148/ryai.2021200171). They hope the model will help earlier detection of cognitive impairment at an individual level.
Brain aging comes with some memory deficits, but determining which impairments are normal and which signal a progression into Alzheimer's disease or other dementias would be extremely useful both therapeutically and for individual care. Amnestic mild cognitive impairment (aMCI) is a transition phase from normal aging to Alzheimer's disease (AD). People with aMCI have memory deficits that are more serious than normal for their age and education, but not severe enough to affect daily function. However for some patients, aMCI is relatively stable while for others it progresses.
A research team based in China sought to determine whether a brain age prediction model based on MRI images could tell the difference between a healthy brain-aging trajectory and the brain-aging seen in patients with aMCI. The model calculated the PAD—predicted age difference—by subtracting the patient’s chronological age from the model’s predicted age based on the images. The team also compared PAD scores with other clinical data, including cognitive impairment, genetic risk factors, pathologic markers of Alzheimer’s Disease (AD), and clinical progression and looked for correlations.
For the retrospective study, the research team trained an elastic net model using brain imaging data from the Beijing Ageing Brain Rejuvenation Initiative (BABRI) and the Alzheimer’s Disease Neuroimaging Initiative (ADNI). The model was trained on T1-weighted MRI images for 974 healthy adults aged 49.3 to 95.4 years. It was then tested on a mix of healthy controls (126 from BABRI and 144 from ADNI) and patients with aMCI (80 from BABRI and 105 from ADNI).
The results showed that aMCI patients had brain-aging trajectories distinct from the typical aging trajectory, and the model could quantify individual deviations from a normal trajectory in these patients. A patient’s PAD score was significantly associated with individual cognitive impairment in several domains, specifically including memory, attention, and executive function.
"The predictive model we generated was highly accurate at estimating chronological age in healthy participants based on only the appearance of MRI scans," the researchers wrote in their paper. "In contrast, for aMCI, the model estimated brain age to be greater than 2.7 years older on average than the patient's chronological age."
The model was also able to distinguish between progressive and stable aMCI. “Combining the predicted age difference with other markers of Alzheimer’s Disease (APOE carrier status, amyloid status, and Mini-Mental Status Examination) showed the highest performance in differentiating progressive aMCI from stable aMCI,” the authors wrote.
They found that progressive aMCI patients exhibit more deviations from typical normal aging than stable aMCI patients, and when combined with other AD-specific biomarkers, the PAD score better predicted the progression of aMCI. For example, Apolipoprotein E (APOE) e4 carriers showed larger predicted age differences than non-carriers, and amyloid-positive patients showed larger predicted age differences than amyloid-negative patients.
"This work indicates that predicted age difference has the potential to be a robust, reliable and computerized biomarker for early diagnosis of cognitive impairment and monitoring response to treatment," the authors concluded.