YouTube Facebook LinkedIn Google+ Twitter Xingrss  

Novel IT Platform Helps Novartis Gain Control of Clinical Imaging Data


ImagEDC’s capabilities are spreading to other trial sponsors.

By Kevin Davies

November 16, 2010 | Researchers at Novartis in Basel have developed a powerful new electronic data capture (EDC) hub for clinical data that has allowed its investigators to control data across multiple trials as never before, according to David Tuch, head of clinical imaging at the Novartis Institute for Biomedical Research, Switzerland. “This is a game changer for managing clinical trials at Novartis.”

Tuch credits his colleague Stefan Baumann, head of imaging infrastructure, for driving the project forward. A physicist by training, Baumann joined Novartis’ clinical imaging team in 2006, charged with managing the imaging IT.

Traditionally, Novartis scientists assessing clinical imaging data would only receive a numeric analysis from a central reading location, rather than the primary image data of the clinical trial, whether MRI, CT, PET, or ultrasound. (Participating hospitals would send their images to a third party for evaluation, which would then send the results to Novartis.) But Novartis investigators wanted the primary image data for a couple of good reasons. First, they wanted to be able to save the image data for later retrieval in exploratory studies with an academic partner. Moreover, access to the primary image data would allow investigators to compare images from earlier trials with improved analysis algorithms.

“Getting ownership around the data—that was my goal,” says Baumann.

The Novartis group built an internal infrastructure as a hub for image data to do quality control. “It’s a perfect infrastructure to analyze and share data with collaborators,” says Baumann. For a start, it is in compliance with Good Clinical Practice (GCP) and Privacy Regulations such as HIPAA and the more stringent European Privacy Regulations. The custom system is based on an Oracle 11g database, using the latest digital imaging and communications in medicine (DICOM) features. Baumann notes that this was a full partnership and his colleagues contributed to the feature set in the database.

Sitting on top of this is an application allowing analysis core labs to log in, submit, and retrieve data. “We can also run quality checking steps and trigger further analysis,” says Baumann. External partners can submit data to the hub, but with a growing number of applicable clinical trials, Novartis needs to take it to the next level to ensure machine-machine integration.

This is where Baumann’s first open-source effort—called ImagEDC—comes in (See: http://code.google.com/p/imagedc/). It allows hospitals and academic core labs to load data, while the tool ensures that the data fit Novartis’ desired format. “It helps any academic partner,” says Baumann. “They don’t need to care about licensing restrictions.”

Not only is it open source and easy to submit data, but it has useful advantages regarding regulatory requirements. “If you’re dealing with clinical trial software, it must be installed and tested in compliance with GCP regulations,” explains Baumann. “Open source enables CROs to integrate code with their own infrastructure. It’s much easier this way.”

Why would Novartis share this software and know-how with its competitors? “The main thinking is that, on the level of transporting images from point A to B, for every trial and hundreds of partners, this is not what we consider to be a competitive advantage,” says Baumann. It’s only later, when one starts to engage in quality checking and analyzing the data, that the business knowledge and competitive advantage comes into play.

Every couple of months, Baumann hosts an informal meeting of a handful of informatics experts at other big pharma companies, where they discuss ways of enabling image data exchange. Work is currently progressing to shape the interface to be broadly usable so other sponsors and academic labs can reuse them. “It’s not just about a tool but specifying a common interface re-usable by everyone,” says Baumann.

Baumann says the success of their infrastructure comes down to the capability to innovate. “We have a handle on data now,” he says. “We can react to what has happened in clinical trials. That’s the big success of the infrastructure.”

We’re Not an IT company

“In 2006, our key objective was to own the data in house,” says Baumann. Of course, handling such large and complex datasets poses some familiar problems. There was concern that clinical trial image data, left in the hands of Novartis’ partners, could end up in a “data grave.” “If storage is not transparent, it could be very hard to make any sense out of the data,” says Baumann. He admits that the scale of the data doesn’t really match that of next-gen sequencing in terms of size, but they are just as complex in terms of the data structure and the distributed nature of clinical trials.

Now, at the push of a button, Baumann’s team can retrieve the images from their central data archive. And with ImagEDC, they’re preparing for a world of federated image data sources. With more and more organizations testing Cloud-based storage options, Baumann says his team is keeping its options open. “We’re not an IT company,” he notes. “Our core knowledge is outside storing Petabytes of data!” His preference would be to find some private clouds to host the data.

Tuch declines to mention the vendor Novartis has used to supply the application layer for the image system, but notes that another pharma is buying a license from the vendor. “We bring in the business knowledge; they offer the finished product to us at a lower price and sell to other parties.”

Baumann offers an anecdote to illustrate the value of their infrastructure. “We had an entity analyzing a multi-center trial, but while it was still ongoing, the entity went bankrupt and couldn’t complete the analysis. That would have been a disaster... But because we had the data loaded into our system, we could a) start the analysis again and b) plugging into this infrastructure, we could plug in a new algorithm to fully automate the analysis and complete that analysis from scratch.”

With ImagEDC, Baumann says, Novartis is using a public platform to transfer data between partners, which involves middleware to transport the data, ensuring high performance and security. “The limit is the bandwidth between the partner and us—there is a bottleneck,” says Baumann. “One option is to use optimized protocols. Then we have a backup option, so if your data density goes beyond a certain point, we use physical media.” •


This article also appeared in the November-December 2010 issue of Bio-IT World Magazine. Subscriptions are free for qualifying individuals. Apply today.


Click here to login and leave a comment.  

0 Comments

Add Comment

Text Only 2000 character limit

Page 1 of 1


For reprints and/or copyright permission, please contact  Terry Manning, 781.972.1349 , tmanning@healthtech.com.