An Indigo Use Case: Studying MRI Images Without Possessing Them
As we work toward AT&T Network 3.0 Indigo, it’s always good to explain what we’re doing. And explain it again.
Indigo will help data-power your world. Here’s how.
It will start with the best qualities of our evolving network, including high speeds and low latency, or lag time. The network must be fast and flexible because huge volumes of data have to move.
Riding the network will be applications – the things that will improve your world with their data power. Maybe it’s a bank’s payment system. Maybe it’s an algorithm for a scientific study. It’s basically software housed on a computer somewhere.
In between the network and the applications will be the guts of Indigo – a platform. If you’re computer-minded, you might call it an abstraction layer. If you’re not, think of a round wooden spool with holes around the outside in a children’s construction set. All the different sticks plug into it. It’s the hub for the spokes.
The spokes include a straw in the cloud where data lives. Maybe it’s a name-brand public cloud, or maybe it’s your own. The straw is secured, and nobody gets access to the straw without multiple identity checks.
Here is an example of how Indigo could help solve a real-world problem:
A hospital system has many locations. Its technicians perform thousands of MRIs at these various locations. The first step in Indigo is to make sure that the right people, and not the wrong people, can access the images easily from any location.
But that’s not enough.
A university researcher wants to develop an algorithm that can detect cancerous tumors earlier than ever before by scanning MRI images. To create, test and improve the algorithm, the researcher needs to train it on thousands of images. But those are private, and they are the property of the hospital system.
A start-up company has a similar idea. It wants to test its idea on those MRI images, too.
The hospital has some hurdles. Patient privacy is one. Also, the MRIs have a dollar value as raw data, but it may be hard to determine. It’d be nice to charge the startup for access, but not give it too much access – certainly not an actual download of the images.
That’s where Indigo comes into play, in something we’re calling a data community. The hospital can own a data community, invite members, and set rules. Community members, like the university researchers and the start-up, might get a detailed description of a data set – in this case, the MRIs. But they wouldn’t have full access to the raw data.
Connectivity and security technologies will guard the community walls, such as those used in AT&T NetBond. Also standing guard will be multi-factor, network-based authentication. Inside, the members can write or execute their own computer code. But, per rules set by the hospital, they are limited on the results they can retrieve.
In this case, they can only get aggregated (grouped) data, not individual records. They can see an MRI image, but not any other data associated with it, like the name of the patient.
All activity will be subject to full and accurate auditing – a key for security and other important factors like academic attribution.
In this example, the hospital has guarded patient privacy, advanced the cure of cancer, and made some money from its data.
Today, people often talk about data in terms like “usefulness versus security,” or “usefulness versus privacy.” What if it became “useful and secure,” and “useful and private?”
That’s the world we’re creating with Indigo.
Victor Nilson - Senior Vice President - Big Data, AT&T