What is it about?
This paper systematically explores new neural network models that aim to detect Alzheimer's disease using MRIs and PET scans. In the comparison, the "fusion" model (depicted in the infographic below) reached a staggering accuracy of 95%. Such a model uses both MRI and PET scans at the same time. Finally, this paper also provides insights into the area of the brains the AI is using to reach such accuracy in the hope of providing valuable data to medical researchers.
Featured Image
Photo by Pawel Czerwinski on Unsplash
Why is it important?
Multi-modal architectures are a new way of dealing with Ai problems, and are thus largely unexplored in the medical diagnosis field. In this paper, they have been revealed to be quite valuable for detecting dementia. An early and accurate detection of this disease may improve the quality of life of patients and their caregivers, as it allows to set up strategies to manage the disease (although a cure does not yet exist). Furthermore, the insights into the AI model inner workings may provide valuable data to the medical research, potentially aiding in the investigation of the disease and the design of a cure.
Perspectives
I loved working on this project. Being able to help advance the field by tackling such an invalidating disease as Alzheimer's allowed me to feel like I was impacting the community - which I hope is true. The explainability results are especially intriguing, as this may guide other researchers to understand the disease better.
Andrea Esposito
Universita degli Studi di Bari Aldo Moro
Read the Original
This page is a summary of: Automated detection of Alzheimer’s disease: a multi-modal approach with 3D MRI and amyloid PET, Scientific Reports, March 2024, Springer Science + Business Media,
DOI: 10.1038/s41598-024-56001-9.
You can read the full text:
Resources
Contributors
The following have contributed to this page
