Why do some children struggle learning to read, and what can be done to help?
A better understanding of the visual portions of children’s brains and how they develop can help provide answers. At Stanford University’s Vision Imaging Science and Technology (VISTA) Lab, Dr. Brian Wandell and his colleagues have analyzed hundreds of magnetic resonance images (MRIs) of different brains. By measuring the developmental changes that occur as reading skills are acquired and correlating neuroimaging data with behavioral data, the lab works to enable more personalized diagnoses and more effective interventions.
VISTA Lab is just one example of many cognitive and neurobiological research projects at Stanford. Most work with large volumes of data and run computationally intensive analyses. When Dr. Wandell became Director of the Stanford Center for Cognitive and Neurobiological Imaging (CNI)—a shared facility that provides imaging and compute resources to more than 40 research labs—he was determined to find a better way to manage, analyze, and share scientific data.
“In the past, we struggled with managing our data, algorithms, and computations, and keeping it all accessible as postdocs come and go,” he says. “We wanted a data management system that would address this challenge while allowing us to share our methods and data with other scientists around the world.”
To make it easier for researchers to organize data and use and share computational tools, VISTA Lab and CNI began using Flywheel, a scientific data management system. The partnership was a success, simplifying the management of neuroimaging data and algorithms.
However, the lab and facility still had to deploy and manage their own hardware, and apply for grant funding when more resources were needed. Procuring and configuring infrastructure for a new project took up to a year. With the growing complexity of neuroimaging and other healthcare-related research, Stanford needed a cloud platform that could handle increasingly complex data computational workloads as well as support HIPAA compliance.
To give its departments a more secure cloud solution for HIPAA-protected information, the university signed a Business Associate Agreement with Google to use Google Cloud Platform. As a result, Google Cloud Platform is now the preferred cloud platform for many Stanford schools and labs, including VISTA Lab, CNI, the Stanford School of Medicine, and the Stanford Center for Genomics and Personalized Medicine. Dr. Wandell and his team had the green light to move Flywheel to the cloud and reap the benefits.
“Running Flywheel on Google Cloud Platform is necessary for our labs to thrive and even survive,” he says. “We can scale easily with research demands, turn resources off when we’re not using them, and extend our labs with a high-performance, highly available cloud that scientists all over the world can tap into.”
With Flywheel on Google Cloud Platform, we can stay closely connected with labs down the street or around the world, greatly increasing the efficiency of our collaborations by simplifying the process of data and computational sharing. We’re helping to keep our researchers focused and productive.Dr. Brian Wandell, Faculty Professor and Director of CNI, Stanford University
Faster time to discovery
Research universities like Stanford spend a significant amount of money on compute and storage resources to process ever-increasing amounts of data. By running virtual machines on Google Compute Engine to manage and scale the Flywheel application, VISTA Lab and CNI are speeding time to discovery while reducing costs.
It’s now much easier for the labs to bring new datasets into the system to produce richer analyses. Researchers can process multiple large datasets in parallel to return results in one day instead of waiting a month or more for serial processing. That agility is empowering researchers like Rosemary Le, a Ph.D. student in psychology studying a portion of the brain critical to the development of reading skills. By bringing in a dataset from Israeli readers and allocating additional compute resources on Google Cloud Platform, Le is accelerating time to discovery for new treatments and interventions.
“With Flywheel on Google Cloud Platform, we can stay closely connected with labs down the street or around the world, greatly increasing the efficiency of our collaborations by simplifying the process of data and computational sharing,” says Dr. Wandell. “We’re helping to keep our researchers focused and productive.”
Previously, the vast majority of IT support time and costs were devoted to managing infrastructure—disk replacements, operating system upgrades, network issues, and system reboots. The new system frees well over half of a full-time employee's (FTE's) time for developing applications.
Adds Michael Perry, Systems Administrator and Software Developer for VISTA Lab: “We’ve seen huge increases in lab productivity by running Flywheel on Google Cloud Platform. Instead of spending half my time managing our compute infrastructure, I can work with our researchers to develop better methods.”
Helping children worldwide
VISTA Lab can now share its valuable datasets and methods with scientists and healthcare providers worldwide, ultimately helping more children get the attention and interventions they need in learning to read.
“By allowing other institutions to access our brain scans, we might be helping a child overcome a learning disability, or enabling earlier and more reliable identification of autism spectrum disorder,” says Dr. Wandell. “Sharing with researchers in Israel and China would have been much harder, if not impossible, without Flywheel on Google Cloud Platform.”
The lab can give its collaborators immediate access to data and methods using their Gmail address, with Flywheel providing identity and rights management for data uploads and downloads. Previously, Michael had to create a Stanford identity for each user and set them up with VPN access—a tedious and time-consuming exercise. As methods change and data is re-analyzed, it’s all available on Google Cloud Platform in real time.
“Scientific collaboration is an area where Flywheel on Google Cloud Platform really shines for us, because we can provide updated data and methods to our collaborators in real time,” says Michael.
We’re saving time and money by running Flywheel on Google Cloud Platform, but what’s most important is the reproducibility we’re able to achieve. The ability to share our research to benefit people all over the world goes right to the heart of science for me.Dr. Brian Wandell, Faculty Professor and Director of CNI, Stanford University
Advancing reproducible science
Stanford is one of the leaders of a movement that is gathering unstoppable momentum. Soon, most, if not all, universities will host large portions of their research infrastructure in the cloud to achieve faster discoveries, reduce costs, and interact with global research communities.
“We’re saving time and money by running Flywheel on Google Cloud Platform, but what’s most important is the reproducibility we’re able to achieve,” says Dr. Wandell. “The ability to share our research to benefit people all over the world goes right to the heart of science for me.”
Adds Travis Richardson, President of Flywheel: “Flywheel’s strategy is to leverage the power of cloud computing to deliver collaborative science while meeting the challenges of massive data and computationally intensive analytics. Our partnerships with Google and Stanford are helping us make this vision a reality.”
We’ve seen huge increases in lab productivity by running Flywheel on Google Cloud Platform. Instead of spending half my time managing our compute infrastructure, I can work with our researchers to develop better methods.Michael Perry, Systems Administrator and Software Developer, Stanford University