According to the World Health Organization the Alzheimer’s disease, the most frequent cause of dementia, affects around 24 million people worldwide, and this number is expected to double in 20 years, resulting in a significant public health burden. Researchers from the University of Kaunas (Lithuania) have developed a model based ondeep learning able to predict the possible onset of the disease from brain images and with an accuracy greater than 99%.
The method achieved better results in terms of accuracy, sensitivity and specificity compared to the previously developed diagnostic tools. The studio was published on “Diagnostics”.
One of the possible first signs of Alzheimer’s is mild cognitive impairment (MCI), which is the intermediate stage between the expected cognitive decline typical of aging and dementia. Other research has shown that functional magnetic resonance imaging (fMRI) can be used to identify brain regions potentially associated with the development of the disease. However, the manual analysis of fMRI images not only does it require specific knowledge, but it takes a lot of time. Furthermore, studying the characteristics of these images does not necessarily indicate the presence of Alzheimer’s, as they may be a symptom of other disorders.
The deep learning model was created by exploiting the potential ofartificial intelligence and a modification of ResNet 18 (residual neural network). The investigation was conducted on functional MRI images of 138 subjects. The latter fell into 6 different categories: from healthy to the spectrum of mild cognitive impairment, to Alzheimer’s disease. In all, 51,443 and 27,310 images were selected from the Alzheimer’s Disease Neuroimaging Initiative fMRI dataset. The model was able to find the characteristics of the MCI with an accuracy of 99.99%.
According to scientists thealgorithm could be developed into software capable of analyzing data collected from vulnerable groups (patients over 65, hypertensive, with a history of brain injury) and thus notify medical staff of the anomalies related to the early onset of Alzheimer’s disease . Finally, this model can be integrated into a more complex system which, taking into account various parameters, would also monitor eye movements, facial expressions, vocal analysis. While they will never replace healthcare professionals, technologies can make medicine more accessible and encourage health early diagnosis.