Alzheimer’s disease (AD) is a progressive neurodegenerative disorder that affects more than five million people in America alone. One of the first signs includes mild cognitive impairment (MCI), which has small variations on the brain changes among intermediates stages and makes early detection difficult for Alzheimer’s since MRI scans cannot pick up alterations before they become extensive enough to be diagnosed as AD itself due to its complexity at this moment.
The race is on to find a faster way of diagnosing Alzheimer’s disease, which affects 24 million people every year and will only get worse as the world’s population ages.
Researchers at the Kaunas University of Technology in Lithuania have developed a deep learning-based method that can predict with over 99% accuracy whether someone has an early stage form or late-stage manifestation for Alzheimer’s disease.
The deep learning-based model developed by the Kaunas University team was a variation on ResNet 18, which classifies fMRI images from 138 subjects. The research team trained their model on workstations equipped with NVIDIA GPUs to examine the effects of Alzheimer’s Disease Neuroimaging Initiative fMRI datasets.
The model effectively found features that distinguished between mild cognitive impairment (MCI) and early Alzheimer’s, achieving a classification accuracy rate of 99.99%. The model also achieved 99.95 percent accuracy when distinguishing between late-stage MCI and Alzheimer’s disease and MCI versus early MCI.
The results published from this study show that the proposed model is more accurate than other known models, with improved sensitivity and specificity.