Insights on Research Data Management

Healthcare research is in a time of rapid change that’s forcing academic researchers to do more with less. Budget constraints often limit the resources available for research, but the demand for insights from imaging data continues to grow. 

To thrive in this environment, you need to adopt strategies that maximize the value of the data you have available without breaking the bank. Here, we’ll review five ways academic researchers can optimize their imaging data even with limited research funding.

1. Share your resources

One of the biggest barriers to efficient data use in research is the siloed nature of information and equipment within institutions. Principal Investigators (PIs) often operate in isolated units, each acquiring and maintaining its own equipment and data. This leads to equipment redundancies and cost inefficiencies. And in times of reduced funding for research, that could mean the difference between success and failure.

That’s why researchers need to embrace the spirit of sharing and collaboration. For example, the University of Colorado Boulder started a project to manage 85 shared instruments utilized by 60 researchers from 18 laboratories, resulting in more than double the seed grant amount in avoided equipment purchase costs alone ($221,000).

Now, imagine what those researchers could do with shared software, repositories and code. 

2. Maximize the potential for systems integration 

Aside from physical siloing, data can be siloed when it is stored in standalone systems that can’t  connect with one another. By utilizing a platform that can integrate with other platforms through open APIs, you get the best of all worlds. 

For example, a radiomics researcher may want to leverage specialized genomics workflows in a genomics offering like Velsera’s CAVATICA cloud-based environment while storing their structured data in an EDC system like RedCap. The power is in being able to connect platforms that excel in different domains and data types.

Integration is critical to maximize the utility of imaging data—and making more of it available to individual research entities. Connecting disparate systems can help researchers unlock previously inaccessible data, drawing connections and insights that enhance the value of their research. 

3. Ensure your data system is versatile and scalable

Data comes in all varieties. In imaging, data formats depend on modality, device and industry standards, among others. Quality of imaging data can also vary wildly, resulting in incomplete data that doesn’t adhere to data standards.

Academic organizations looking to improve imaging data management should consider a data type-agnostic platform to serve all divisions of an institution. Some purpose-built data platforms can handle various types of data, allowing institutions to funnel all of their data into one location. 

Reimagining entire systems from the ground up to allow for integration of multiple systems can be overly costly in terms of time, money and effort. But a platform that can sit seamlessly on top of a global cloud or hybrid infrastructure doesn’t require a complete overhaul. This approach not only saves on the costs associated with acquiring new infrastructure but also enhances data accessibility and usability.

4. Make big data: focus on data aggregation

Data at academic medical organizations can be a gold mine. But you can’t use what you don’t know is there or what you can’t see.

A vast majority of academic research data is underutilized simply because researchers are unaware of its existence or location. One estimate by GE Healthcare puts it at 97% of valuable healthcare data that goes underutilized. With 90% of healthcare data coming from imaging, that’s a huge wealth of data we’re just sitting on that could go toward advancing lifechanging healthcare initiatives. In fact, the McKinsey Global Institute estimates that applying big-data strategies could generate up to $100 billion annually in the U.S. healthcare system, in part by improving the efficiency of research and clinical trials.

Aggregating data is the first step to prioritizing data discovery. It provides a comprehensive view of an institution’s data footprint, allowing researchers to build more powerful cohorts, driving more robust conclusions and insights. This approach not only maximizes the utility of existing data but also avoids unnecessary spending to acquire new data.

5. Leverage AI to speed time to Insights

AI has already been proven to enhance data analysis and interpretation by dramatically augmenting human capabilities, especially in the medical imaging space In fact, of the almost 700 FDA-approved Artificial Intelligence and Machine Learning (AI/ML)-Enabled Medical Devices, 77% of them are for radiological (medical imaging) applications.

To develop AI algorithms that work for your needs, you need good, clean and analysis-ready data. For instance, a research team at University of Wisconsin–Madison worked remotely to migrate 50,000 datasets of chest X-rays and PCR test results from various sources into the Flywheel medical imaging platform to develop an AI model that achieved a diagnosis accuracy of 94%. 

The result? The model outperformed experienced thoracic radiologists by 9%, and they were able to reduce manual data processing work from eight months down to one day.

Putting It All Together in One Platform

Now is the time to maximize the value of your imaging data while minimizing cost. Achieve this by adopting a collaborative platform that integrates multiple systems, can traverse existing and cloud infrastructure for rapid shift and scale, fosters data aggregation for discovery, and allows for AI algorithm implementation and creation of datasets for AI development. 

One platform that embodies these principles is Flywheel. We’ve worked with folks who cut their teeth in both academic and pharmaceutical research to develop an all-encompassing solution designed to meet the needs of modern researchers, allowing you to accelerate important research initiatives and gain more visibility into your data—all while staying within increasingly tight budgets and timelines.

To learn more about how Flywheel medical imaging can assist in optimizing your research with constrained resources, make sure to attend our session, “One Platform to Rule Them All: Empowering Multimodal Imaging & Video Analysis & AI” at the SIIM25 Annual Meeting + InformaticsTECH Expo, taking place in the Expo Hall May 21 from 1-1:20 p.m. PT. We’ll also be at attendance at the show at booth #232. Stop by, and start revolutionizing your research capabilities.

Erin Chu, DVM, PhD, is Director of Technical Sales at Flywheel, where she helps researchers optimize their medical imaging data for insights. Prior to joining Flywheel, Erin was the global lead for Amazon Web Services Open Data, where she helped democratize access to petabytes of high value datasets including the Imaging Data Commons and the Sequence Read Archive.