The application of artificial intelligence and machine learning (AI/ML) in medical imaging has shown remarkable effectiveness. The FDA has sanctioned nearly 700 AI/ML-Enabled Medical Devices, with 77% tailored for radiological uses. Medical imaging plays an essential role in diagnostic processes. In 2023, the United States recorded around 500 million medical imaging exams for adults, averaging nearly two exams per individual. Moreover, most states mandate medical data retention for a minimum of five years. Given the time-consuming nature of medical imaging, radiology technicians and radiologists require extensive training to capture and interpret these examinations. While a basic X-ray may take less than a minute, more complex procedures like Magnetic Resonance Imaging (MRI) can extend up to an hour.
Healthcare providers encounter several significant obstacles in utilizing medical imaging data for machine learning applications. One major challenge is the fragmented storage of data—medical imaging information often resides in various disconnected repositories within healthcare systems. Additionally, strict compliance regulations necessitate that organizations meticulously preprocess their imaging data, implementing rigorous deidentification, standardization, and harmonization workflows prior to integrating it into ML models. The vast scale of medical imaging data further exacerbates these issues, with hospitals generating an average of 50 petabytes of data annually, of which around 80% consists of medical images. Hence, organizations require solutions capable of efficiently and securely managing large datasets.
Flywheel Solution Overview
Flywheel operates on AWS to deliver a Software as a Service (SaaS) model. As illustrated in Figure 1, Flywheel leverages AWS services to facilitate complete processing of medical imaging data at scale. It integrates seamlessly with Amazon Simple Storage Service (Amazon S3) for secure, scalable storage and intelligent cost optimization.
This integration allows healthcare organizations to manage, analyze, and share medical imaging data while simultaneously cutting down on costs and time required for data preparation. AWS customers have reported achieving up to a 94% reduction in medical imaging workflow costs, transforming previously months-long manual processes into mere days.
Amazon S3 provides effortless scalability, ensuring performance while accommodating growing data volumes. Security remains a top priority for health data; Amazon S3 is HIPAA eligible and boasts enhanced encryption features, access controls, and audit logging to maintain data integrity and comply with regulations. Utilizing Flywheel’s Interfaces tab, users can easily mount multiple Amazon S3 buckets for data import, export, or both with a click of a button.
To secure access to your medical imaging data, you can easily link your Amazon S3 bucket to Flywheel by configuring AWS credentials. Begin by designating your S3 bucket (e.g., s3://my-flywheel-bucket) within the Flywheel interface. Then, input the AWS Access Key and Secret Access Key associated with an AWS Identity and Access Management (IAM) role that grants the necessary permissions to access the bucket. This setup, as shown in Figure 2, creates a secure connection between Flywheel and your S3 storage.
Once your S3 bucket is mounted, importing data into a Flywheel Project is straightforward using the Web Import tool, as demonstrated in Figure 3. For instance, we imported data regarding one subject from the NSCLC-Radiomics-Interobserver1 project, which is part of the Imaging Data Commons and freely available via the Registry of Open Data on AWS.
Within a Flywheel project, data is organized by Subject, Session, Acquisition, and File, as depicted in Figure 4. Flywheel simplifies data management by consolidating various data types in one location. For example, a patient may undergo an X-Ray one visit, followed by an MRI in the next, with survey results saved as a PDF or CSV; Flywheel categorizes all these data types under the same Subject.
The platform allows users to visualize data using Flywheel’s zero-footprint radiology viewer, enabling annotations and segmentation of medical images. Flywheel also optimizes storage costs by transitioning data between access tiers based on usage patterns through Amazon S3 Intelligent Tiering. By employing the Reference-in-Place feature, Flywheel avoids duplicating files, which cuts down on storage expenses. Instead, it reads the source files to extract metadata while storing a reference, treating the source as the primary storage.
Preparing Data for Analysis with Flywheel on AWS
After importing data into Flywheel, you can process and analyze it using Flywheel Gears, which utilize configurable Amazon Elastic Compute Cloud (Amazon EC2) instances for maximizing time and cost efficiency. If your Project has a Flywheel Gear Rule enabled, you can also trigger automatic data pre-processing. Our Example Project implements a simple Gear Rule that extracts Digital Imaging and Communications in Medicine (DICOM) header metadata and functionally classifies incoming DICOM files. The outcomes of these processes are displayed in the Session dashboard shown in Figure 3.
Additionally, machine learning models can be trained directly within the Flywheel platform using Flywheel Workspaces. Cohort building is made easy with Flywheel’s Search function, allowing searches based on medical image attributes or accompanying structured data.
Sharing data with Flywheel is uncomplicated. Collaborators can access data through Flywheel’s granular roles and permissions or via the Flywheel SmartCopy feature. Alternatively, exporting a Flywheel project to an Amazon S3 bucket for retrieval is also a viable option.
By incorporating Flywheel into your AWS solutions, you can minimize costs and time spent on the tedious tasks of data aggregation, processing, and curation, allowing for a greater focus on generating insights. For further reading, visit this blog post to explore more on this topic.
Empowering Healthcare with Flywheel and Amazon S3
The partnership between Flywheel and Amazon S3 presents a powerful solution for optimizing medical imaging workflows. This integration enables healthcare organizations to effectively manage, analyze, and share extensive imaging data while significantly reducing costs and preparation time. As the amount of medical imaging data continues to escalate, Flywheel on AWS delivers a robust, scalable framework for promoting AI-driven healthcare advancements. For those interested in a detailed resource, check out this page, as they are an authority on this topic.
If you want to learn more about how Flywheel can enhance your medical imaging and associated health data, visit flywheel.io. Moreover, for insights into onboarding experiences, you can refer to this excellent resource.
Leave a Reply