How GPUs Are Powering the Science of Memory
What does a gaming class GPU have to do with understanding human memory and brain disease? Quite a lot. A research team working at the Marine Biological Laboratory in Woods Hole, Massachusetts is using NVIDIA RTX GPUs and HP Z workstations to explore how our brains store memories at the molecular level.
Their work focuses on the hippocampus, a seahorse shaped part of the brain that plays a key role in forming and storing memories. Using powerful GPUs, high performance workstations and virtual reality tools, the team is turning massive brain imaging datasets into interactive 3D environments that scientists and even high school students can explore.
This project shows how hardware that comes from the same GPU families used for gaming and content creation can also drive cutting edge scientific discovery and visualization.
Inside the Neural Forest: Massive 3D Brain Data
To understand memory, the researchers zoom in far beyond what the naked eye can see. They are studying brain cells in the hippocampus at extremely high resolution, focusing on tiny protein markers that may be involved in how memories are encoded.
The lead scientist, Andre Fenton from New York University, describes the hippocampus like a dense forest. Each neuron is like a tree trunk and its branching structures are like leaves. Hidden among this forest are specific protein markers that matter for memory, and they make up only about one percent of all such markers in the region. Finding and studying them is like searching for a few special leaves in a forest of billions of trees.
To capture this detail, the team collects huge volumes of 3D volumetric data. With traditional tools, simply storing and checking these images was a bottleneck. Once NVIDIA RTX GPUs and HP Z workstations were integrated into the workflow, everything changed.
- The researchers captured around 10 terabytes of high resolution 3D brain data.
- NVIDIA RTX GPUs handle the heavy lifting of rendering and processing this data.
- HP Z high performance workstations provide the CPU power, memory and storage needed to manage and inspect the massive datasets.
The GPU acceleration makes it possible to perform visual quality checks and interactive exploration of the data that would otherwise be painfully slow or even impossible on typical desktop systems.
On the software side, the team uses syGlass, a virtual reality platform built for scientific visualization. Running on HP Z6 desktop workstations with multiple NVIDIA RTX GPUs, syGlass allows the researchers to step inside the 3D brain data in virtual reality and inspect structures in real time.
From GPUs to Breakthroughs in Brain Health
The scientific goal behind all this tech is serious. The team wants to understand memory at a molecular level. By learning how proteins are organized in the hippocampus and how their location and structure relate to memory function, scientists may uncover clues about diseases such as Alzheimer’s and dementia.
Researchers are particularly interested in what happens when these key proteins end up in the wrong place. By comparing healthy patterns to altered ones, they hope to reveal how memory fails and how those failures connect to neuropsychiatric and neurocognitive disorders.
All of this requires extremely detailed, high resolution 3D images and the ability to navigate them without waiting minutes for each frame to render. This is exactly where GPUs shine. The same parallel processing power that boosts frame rates in modern PC games now helps scientists fly through dense brain data smoothly in VR, checking structures from every angle and marking important protein markers.
As co investigator Abhishek Kumar points out, once you know how something is built, you have a much better chance of understanding what goes wrong and how to fix it. For the brain, that could mean new insights into how we form memories and how to treat conditions where memory and cognition start to break down.
VR Headsets, High School Students and Hands On Science
One of the most interesting twists in this project is how accessible it becomes once VR is in the mix. Using syGlass in combination with NVIDIA RTX powered HP Z workstations, the team turned a tedious research problem into an engaging virtual exploration.
They brought three high school interns into the lab and gave them VR headsets connected to the GPU accelerated systems. Inside virtual reality, the students could see the hippocampal data in full 3D, move through the neuron forest and search for specific protein markers related to memory.
Their job was to identify and label the correct proteins in a sea of billions of neurons and countless markers, a task that is challenging but surprisingly well suited to human pattern recognition when given the right tools. Thanks to real time rendering on RTX GPUs, the students could interact smoothly with the data rather than waiting for slow refreshes or struggling with lag.
The pilot program was a success. The researchers are now considering scaling this approach to include more students at multiple locations. With the right mix of hardware, software and VR, complex brain research becomes something that motivated students can directly contribute to, while also learning about neuroscience and advanced computing.
This project is a good example of how PC class hardware like NVIDIA RTX GPUs and HP Z workstations can cross over from gaming and creative workloads into serious scientific visualization. The same performance boosts that gamers look for in higher frame rates and smoother graphics are exactly what allow scientists to explore massive 3D datasets interactively, turning raw data into new understanding of how our brains work.
Original article and image: https://blogs.nvidia.com/blog/mbl-human-memory-ai-vr-rtx/
