If you thought studying Electrical and Computer Engineering prevents you from learning about the human body, think again. Dr. Eva Dyer, a BME Assistant Professor at Georgia Tech, uses her ECE education to study the brain in the Neural Data Science (NeRDS) lab that she runs.
The brain is made up of millions of neurons, forming different regions that control everything from memory storage to hand movements. One challenge with studying an organ as complex as the brain is creating detailed images of its different regions and the connections between them.
One such way of creating brain images is by doing functional MRI (fMRI) scans. This measures brain activity by detecting the changes in blood flow to various regions of the brain.
Electron microscopy, on the other hand, allows scientists to scan thin slices of brain sample and see each neuron and connecting blood vessels. This provides a zoomed-in view of one specific slice of the brain but does not show how different areas of the brain interact with each other.
Dr. Dyer investigated the use of x-ray microtomography to produce brain images which involves shooting x-rays at a brain sample and capturing the reflected x-rays. These reflected waves are converted into visible light using a crystal before being photographed by a camera. By rotating the sample as it is exposed to x-rays, a 3-D image of the sample is reconstructed. This scan used high-energy x-ray photons generated in a synchrotron at the Argonne National Lab (ANL), a facility the size of dozens of football fields.
In a proof of concept study, Dr. Dyer showed x-ray microtomography can take a large brain sample (1 mm thick) and produce an image with micrometer-level resolution. Another benefit of this method is its integration with an algorithm she created that can distinguish between blood vessels and cell bodies in the image produced.
In order to implement this method to research and clinical applications, her team is looking into how to scale up the process so that larger brain samples can be imaged without losing resolution or damaging the sample with strong x-rays. In addition, integration of microtomography results with existing techniques such as electron microscopy and fMRI can give researchers the ability to transition seamlessly between a high-level general overview of brain activity and a detailed map of a specific region.
Dr. Dyer is also interested in looking at how neural patterns can be used to decode how our limbs move. Currently, brain-computer interfaces (such as those that control prosthetic limbs) use an algorithm called supervised decoder. This algorithm uses simultaneous recordings of brain patterns and moment-by-moment movement details to teach itself how to translate brain patterns into limb movements. This is both time consuming and labor intensive, which slows down the progress of brain-controlled prosthetics.
In another proof of concept study, she used brain pattern recordings combined with a general understanding of the brain patterns that pop up when a limb moves in a certain way to predict what movement the limb would make.
The team then created a model to match the monkeys’ brain patterns with their actual hand movements. The best model they created was able to predict hand movements about as well as basic supervised decoders already in use, and in the future they hope to improve the model accuracy even further.
She has also published some of the tools she developed for her various projects for other research groups to use, including the algorithm she created to annotate x-ray microtomography images of the brain. To find out more about her work and other projects she has in the pipeline, you can check out her website at DyerLab.gatech.edu.