Scientists have developed a new computer-based method that shows great promise in improving the accuracy of diagnosis of neurodegenerative diseases, including Alzheimer’s and Parkinson’s disease.
The study, “Ensembles of Deep Learning Architectures for the Early Diagnosis of the Alzheimer’s Disease,” was published in the International Journal Of Neural Systems journal.
Despite the substantial progresses in therapeutics targeting both Parkinson’s and Alzheimer’s, the diseases remains without a cure. Early diagnosis is crucial for the success of existing therapies in improving patients’ lives, as well as to help develop new medicines.
Computer aided diagnosis (CAD) has been very helpful in helping physicians to determine brain functions by analyzing the multimedia results obtained in tests carried out in patients. Imaging the brain is one of such tools, as it allows physicians to visualize patients responses in real time.
These types of biomedical images have been the study focus of the research team at the University of Malaga and collaborators at the University of Granada.
The team developed a new method based on what is known as deep learning architectures. The self-learning algorithms that composed this method identified brain regions and connections that were abnormal in Alzheimer’s disease. It also identified and extracted the most relevant characteristics of several images.
In simpler terms, researchers can use these deep learning architectures to create and modulate complex images. The team can fuse two different types of images — functional and structural — to create an atlas of the human brain. This method allows researchers to compare structural changes in the brain’s gray matter with functional alterations in white matter (nerve fibers that connect different brain areas).
“The study uses deep learning techniques to calculate brain function predictors and magnetic resonance imaging to prevent Alzheimer’s disease. To do this, we have used different neural networks with which to model each region of the brain to combine them afterward,” researchers explained in a press release.
This new method has been validated using a large dataset from the Alzheimer’s Disease Neuroimaging Initiative (ADNI), which showed that the method is not only valid for differentiating “between controls and Alzheimer’s disease images, but it also provides good performances when tested for the more challenging cases of classifying mild cognitive impairment (MCI) subjects.”
Overall, these results suggest that this new method will not only contribute to a deeper understanding of neurodegenerative diseases, but may also help the development of more efficient therapeutics. For now, this method promises to substantially improve the diagnosis of other diseases, including Parkinson’s disease.