Flywheel Enterprise

I have been actively using Flywheel.io for the last two years.

My main use case for Flywheel.io ranges from data ingestion to algorithm development, and primarily for the last year, we have been using it for standardizing our data and improving our ingestion pipelines.

A specific example of how I have used Flywheel.io for improving our ingestion pipelines is that we have used multiple gears, which allow us to standardize our data during ingestion. These gears range from DICOM-based ingestion to NIfTI-based ingestion, and we have some QC gears and some metadata extraction gears that we run in a chain to improve our data ingestion.

Apart from this, we also use Flywheel.io for some of our algorithm development. For example, we use it for removing text from images or for concatenating images, and there are multiple use cases in our company that we run on Flywheel.io.

Based on our experience, the best features Flywheel.io offers are its great flexibility to run the gears and interact with Flywheel.io. For example, it offers an SDK and the CLI, which makes it a very complete product.

The SDK and CLI have been helpful for us because we primarily use the SDK, and we use the CLI for local testing of our gears. While we are developing them, it really helps us to debug issues and speed up the development process. The SDK is always useful when we want to interact with or search data in our projects that contain a very large amount of data. The SDK really helps us to understand our data or to identify any problems with the data.

Flywheel.io has positively impacted our organization because it helps us to standardize our data coming from different CROs, which is organized in different ways. The most important benefit we receive from Flywheel.io is that it ensures our data is ready for algorithm analysis once it goes through the checks in Flywheel.io.

While it is hard to quantify in a measurable sense, it saves the time that would be required if the data were not standardized. Without standardization, developing our AI-based models would require approximately 70 to 80 percent of the time needed just to clean up the data, and that time is saved when the data is in proper formats.

One way Flywheel.io needs to improve is by having more availability of computational power required in this time of AI. It needs to understand those requirements dynamically and should use GPU-intensive models wherever they are available to speed up the process. Secondly, when we run multiple jobs, it sometimes becomes a difficult and slow process, especially for larger algorithms.

The user interface on Flywheel.io's side is quite good, and I would not add anything there.

Regarding Flywheel.io's AI capabilities, I think the governance part is still laggy from our experience and needs improvement. The security part is somewhat good.

Flywheel.io is quite stable.

Flywheel.io scales well. The only issue I have experienced is that when there are a lot of jobs, it sometimes feels laggy because running many jobs can block the system overall for other users, which affects scalability.

We always receive good support, and the customer support is fantastic. I rate the customer support at 10.

While it is hard to quantify in a measurable sense, it saves the time that would be required if the data were not standardized. Without standardization, developing our AI-based models would require approximately 70 to 80 percent of the time needed just to clean up the data, and that time is saved when the data is in proper formats.

I would certainly advise others to use Flywheel.io for medical imaging tasks.

Flywheel.io is deployed in our organization as a hybrid cloud system.

We use multiple cloud providers, including Amazon Web Services and Google Cloud Platform, which are all three main players. I rate this review 8 out of 10.