Accessible brain scanning for Alzheimer’s disease

Alzheimer’s disease (AD) is the most common cause of dementia in the world and research has been able to identify brain changes associated with the degenerative process of the disease. Moreover, we now know that these changes might be observed in brain scans at the onset of the mildest symptoms or even years before. Even though current treatments are somewhat limited, one thing has become clear, the sooner treatment is started the more effective it is. The problem is that brain scanning is not as easy – or cheap – as it might sound. Because a magnetic resonance imaging (MRI) machine needs a giant magnet to produce brain images, it might weigh between 22,000 to 33,000lbs (10,000-15,000kgs) and cost around 3 million USD. As you can probably imagine, the infrastructure and funding needed to maintain such machines is no easy feat in public health systems.

How then do we make brain scanning for AD more accessible? Nature Communications recently published an amazing study done by Sorby-Adams and colleagues where they tested the feasibility of using portable, low-field magnetic resonance imaging (LF-MRI). This new technology has less power and infrastructure requirements and therefore costs much less. However, there is a price to pay for paying less. Because the magnet is not as big, its magnetic field is not as strong. This causes the produced images to have lower values of a quality measure called signal-to-noise ratio (SNR), which affects the clarity and resolution of the images. The authors of the study developed machine learning tools to work around these resolution limitations and still be able to identify brain changes in patients with mild cognitive impairment (MCI) and AD using LF-MRI. 

The study analyzed data from people that had been scanned both with a portable LF-MRI scanner and with a conventional (i.e., high field) MRI scanner. The images obtained from the latter scanner were considered the standard or ground truth. Scanned participants included healthy individuals, those with cardiovascular risks but no signs of memory problems, and patients with mild cognitive impairment (MCI) or AD. The variability in the sample helped to get a good spectrum of different clinical stages and therefore obtain more reliable results. LF-SynthSR is the algorithm they developed to enhance LF-MRI scans. This tool uses artificial intelligence powered by a special neural network called a U-net. It works with an initial training phase where fake LF scans that are generated are then compared with the high field images (i.e., ground truth). The algorithm learns by comparing differences in image intensity and how well the images are segmented into specific brain regions. This process teaches the system how to “translate” LF scans into high-quality outputs.  Then, the real LF-MRI scans collected from participants are enhanced in order to be properly analyzed for dementia-related signatures. 

To summarize the results, it worked. The new streamlined process tested by the authors was able to obtain accurate brain measurements out of LF processed scans. These measurements included the hippocampus volume (a key brain region for memory) and white matter lesions (areas of brain damage frequently found in AD). Moreover, when applied to MCI and AD patients the pipeline revealed patterns of brain shrinkage and white matter changes similar to those seen in traditional HF-MRI. 

This study shows that portable LF-MRI scans, when paired with advanced automated analysis, can provide reliable brain measurements comparable to HF-MRI. It seems that LF-MRI tools have the potential to distinguish between patients with MCI or AD and those without it, making it a promising, cost-effective tool for broader more accessible clinical use. It is exciting to think that we might be getting closer to making early diagnosis of Alzheimer’s accessible to everyone, regardless of where they live. 

Source: 

Sorby-Adams, A.J., Guo, J., Laso, P. et al. Portable, low-field magnetic resonance imaging for evaluation of Alzheimer's disease. Nat Commun 15, 10488 (2024). https://doi.org/10.1038/s41467-024-54972-x

Brain Bites Sub-Team
Lead: Alejandra Lopez-Castro