A Combination Single Molecule Microscope and ISIM (Instant Structured Illumination Microscope) in the AIM facility. Photo credit: Harshad Vishwasrao, AIM Managing DirectorEvery day, vast amounts of imaging data are produced through medical checkups, screenings, and procedures—eye exams, X-rays, MRIs, CT scans, ultrasounds, and so on. This raw imaging data paired with powerful computational tools and techniques available in the cloud—such as machine learning and deep learning—allow for 3D models that help researchers at the National Institutes of Health (NIH) identify abnormalities that indicate diseases. All with the hope of aiding the development of possible treatments and therapeutics for people whose lives are impacted by these conditions.
NIH’s Advanced Imaging and Microscopy (AIM) resource—housed on the main campus in Bethesda, MD—is home to cutting-edge, NIH-developed microscopes that produce imaging data for research areas such as ophthalmology and neuroscience. Staffed with experts in microscope technology and computational imaging, the AIM resource provides NIH intramural researchers a core facility to use the advanced imaging microscopes without the need to be an expert in the technology.
Like many on-site resources, there are capacity challenges around availability of microscopes, computational resources, and data storage. “Getting to the cloud is a really important step for the long-term,” explained Johnny Tam, Ph.D., a Stadtman Tenure-track Investigator at the National Eye Institute (NEI) and steering committee member for the AIM resource. “The cloud is scalable and can help alleviate bottlenecks in data collection and data analysis.”
AIM has only just begun its cloud journey, but some of its users have already seen great promise in translating their programs to the cloud. “We were able to use the funds from the NIH Office of Data Science Strategy to access the cloud [through the STRIDES Initiative] and crunch the data much faster than we could using the local hardware in our labs. The impact of this was that within a couple of weeks, we could complete what would have taken us maybe months locally,” said Hari Shroff, Ph.D., a principal investigator at the National Institute of Biomedical Imaging and Bioengineering (NIBIB) and lead of the AIM resource.
Leveraging the STRIDES Initiative
Similar to other new tools and systems, there is a learning curve to using the cloud for imaging data. The STRIDES Initiative has helped to bridge the gap for NEI, NIBIB, and the AIM resource by providing access to the tools, training, and experts they needed to understand how to get set up in the cloud. “We see many parallels between the AIM resource and the STRIDES Initiative. When we began looking into cloud computing for our research, we didn’t have much expertise. The STRIDES Initiative gave us the impetus to get started with the cloud,” said Tam.
“We are grateful for the opportunity to use the STRIDES Initiative as it has opened our eyes to things we had not thought about. With [imaging-based] research, you are consistently thinking about what you can do next to further your research and possibly open doors to a new field of study or instrument. This engagement [with the STRIDES Initiative] is going to allow us to do that.” Johnny Tam, PhD, Stadtman Tenure-track Investigator at the National Eye Institute, and steering committee member for the AIM resource
Impact of the Cloud on Imaging Data
AIM, in collaboration with NEI and NIBIB, began exploring how the cloud could help democratize access to the non-commercial microscopes and expedite analyses of imaging data for the more than 5,200 scientists and fellows in the NIH Intramural Research Program and possibly beyond NIH.
In general, the imaging data produced by the microscopes are degraded in some way but can be enhanced using techniques like machine learning and deep learning. However, it takes time to enhance the images using these computational analyses on local infrastructure. With the scalable resources available in the cloud, NEI, NIBIB, and AIM can analyze the data significantly faster.
Leveraging the STRIDES Initiative to access cloud tools and services led to:
- Getting started with cloud resources more quickly than doing it alone
- A decrease in data processing time
- The ability to refine and retrain models to use on new studies
- The ability to run multiple studies at once
- A scaling of resources to support more researchers using AIM
- Seamless use of data without the need to transfer it from different computers
- Access to cloud computational tools for labs that wouldn’t have it otherwise
- The ability to complete studies and publish papers faster
Challenges
- Scaling resources
- Analyzing data quickly
- Reusing existing models
- Understanding the cloud
Key Results
- Democratizing access to resources
- Expediting research analyses
- Lowering the barrier to entry
- Expanding expertise for using the cloud
Future Plans
Medical imaging data analysis is starting to become more popular in the cloud and is a key topic for the computer science and medical imaging analysis community. With imaging data, researchers can more intelligently make decisions about how to image and where to source the imaging data from so that they can more accurately conduct research studies. Tam explained that there are many opportunities to incorporate the use of cloud—from a more efficient computing environment and robust machine learning to quickly interpreting data—ultimately offering faster alternatives to local infrastructure.
Currently, more than 30 research groups from almost every NIH Institute and Center leverage the microscopes in AIM annually—imaging everything from dynamic organelles within living cells to entirely fixed mouse brains. In the future, the goal for AIM is to increase the availability of the microscopes and technology for more researchers at NIH. The cloud offers the ability to streamline support locally, and the STRIDES Initiative reduces the barrier for researchers who want to use the technology. Today’s work to create standards of cloud services for imaging data will give future generations a starting point to move imaging in the cloud forward.
Description of the Program
The National Institute of Biomedical Imaging and Bioengineering (NIBIB), the National Eye Institute (NEI) work in collaboration with Advanced Imaging and Microscopy (AIM), a trans-NIH resource, that houses, operates, disseminates, and improves non-commercial, prototype microscopes developed at the NIH. The resources include five cutting-edge microscopes and are available to the entire NIH Intramural Research Program.
