Thursday, November 12, 2009

"The Only FDA Approved Remote Review System"

You might recall the "shootout" between Visage and TeraRecon we held here a few months ago. GE could not attend due to timing conflicts, but later held a separate demonstration of their AW Server system at our site. In brief, this is a thin-client version of the venerable AW software, which allows remote access to high-level processing capabilities. The system works well, especially for those who like the GE/AW approach to such things. It has a lot of power, but doesn't quite have the level of automation that was demonstrated by TeraRecon.

Our GE rep, of whom I'm far more of a fan than of the company itself, sent us these bullet points about the AW Server:

Designed with you…

  • Enhanced work flow tools for ultimate reading and collaboration
  • Easily track workflow status with custom work lists and filters
  • Switch between exams instantly to minimize interruptions
  • Easy access to all prior exams in the work list for quick comparisons

Beyond the walls…..

Diagnostic reads with Exclusive “Smart Compression” technology….Accessibility anywhere, anytime

  • Removes constraints on needing physical access to a workstation…Convert any PC to a high end, post processing workstation
  • Total thin Client solution
  • Only FDA approved Remote Review System
  • Unique “Smart Compression” Technology
  • One click to apply compression only during interaction
  • Compression “smartly” turns off when the mouse is released
  • Automatically display full fidelity static, high quality images for diagnosis!
  • Works over any network condition…from any location

Secure the future…

Seamless integration with the security and service features facilities need…Maximize investment

Of these claims, clearly the statement "Only FDA approved Remote Review System" catches the eye. I did a double-take on this, both when I read the e-mail, and when it was stated in the demonstration. This seems to be a rather bold statement on the surface, and I thought I should probably pursue it further. So, whilst at RANZCR, I asked the president of Promedicus, the company that owns Visage, if he was aware of this. He thought the statement was incorrect, and assumed that Visage, TeraRecon, and other similar products were all approved in the U.S. Curiouser and curiouser.

There was nothing left to do but ask GE directly, which I did via email. The answer came with an attached market clearance document from the FDA, although not quite the one that is found on the FDA website. The publicly available document is found HERE, and differs from the document GE sent me in its opening pages. The FDA's version has some different cover-sheets which we will discuss in a minute. The salient point from GE is found on the last page:


The device is not intended for diagnosis of mammography images. The device is not intended for diagnosis of lossy compressed images. For other images, trained physicians may use the images as a basis for diagnosis upon ensuring that monitor quality, ambient light conditions and image compression ratios are consistent with clinical application.

Italics are mine. The interpretation of this line is that one should only be diagnosing from full-fidelity images, and not lossy-compressed images. (Lossless compressed images are a no man's land, I guess.) GE's approach is something called "Smart Compression" as seen in their AW Server brochure below on page 4. (Sorry about the Scribd embedding, but I could not find the same brochure online.)



The important line reads:

Exclusive “Smart Compression” technology dynamically speeds you through the case – even at low bandwidths – so you can quickly view full fidelity, static images for instant diagnoses.
All this was reviewed in a conference call with several high-level folks in the AW division of GE Medical. This is the technology upon which they base the claim.

The approach is somewhat reminiscent of streaming technology as seen in Stentor and Dynamic Imaging PACS wherein the compressed images are around when you interact with the study, scroll, move around, whatever, but the full-fidelity image is displayed when you let up on the mouse. I assume GE's software decompresses lossless-compressed images on-the-fly to satisfy the claim of presenting full-fidelity images. How this applies to a 3D volume rendering or a MIP series or whatever, I haven't a clue.

The question now becomes, "What about the other guys?" What I'm going to do here is throw it open to any other vendor in this space that wants to respond with something I can post, and I will post it without editing or editorializing. I'm expecting responses from TeraRecon and Visage at the very least. Are you guys listening?

In the meantime, I did some additional digging over at the FDA.gov website. First, let's look at TeraRecon's 510(k) Premarket Notification, from December, 2005. Here are its indications for use:

The AquariusAPS server receives medical images from medical imaging acquisition devices adhering to the DICOM protocol for image transfer such as EBT, CT, MRI, and other volumetric or planar medical imaging modalities, and performs digital image processing to derive certain information or new images from these image sets. The information or new images thus derived is transmitted using the DICOM protocol to other devices supporting this standard protocol.

Lossy compressed mammographic images and digitized film screen images must not be reviewed for primary image interpretations. Mammographic images may only be interpreted using an FDA approved monitor that offers at least 5 MP resolution and meets other technical specifications reviewed and accepted by FDA." The intended use of the device is to provide time-saving pre-processing of images to remove the need for an image review system to perform these activities while a user is waiting for processing to complete, to optimize the use of the user's time.

The issue of using the thing for diagnosis really isn't addressed directly; one assumes that the FDA's permission to sell the device means we radiologists can actually use the device, I guess. And, they only discuss compression with respect to digital mammography.

When we look at the Visage 510(k) paperwork, things start to get a little murkier. Visage applied for approval for its PACS and CS (advanced visualization) software in one bundle.

Intended Use

Visage PACS/CS is a system for distributing, viewing, processing, and archiving medical images within and outside health care environments.

The Visage PACS/CS server receives image data in DICOM format via the hospital network. This provides universal connections to archives, modalities, and workstations.


Visage PACS/CS is to be used only by trained and instructed health care professionals. It can support physicians and/or their medical staff in providing their own diagnosis for medical cases. The final decision regarding diagnoses, however, resides with the doctors and/or their medical staff in their own area of responsibility.

Although the web and thin client technologies allow the software to be run on a variety of hardware platforms, for diagnostic purposes the user must make sure that the display hardware used for reading the images complies with state-of-the-art diagnostic requirements and currently valid laws.

Only DICOM for presentation images can be used on an FDA approved monitor for mammography for primary image diagnosis.

Only uncompressed or non-lossy compressed images must be used for primary image diagnosis in mammography.

It sounds to me like Visage has the same approval level. By the way, I've come to the realization that the indications here, as for the AW and TeraRecon, are written by the companies themselves, and approved by the FDA. That does help somewhat in the interpretation, doesn't it?

Which brings us back to the cover letter of the AW Server 510(k) application from the FDA. It includes the following:


AW Server is substantially equivalent to the devices listed below:

Model: Advantage Workstation 4.3
Manufacturer: General Electric Medical Systems
510 (k): K052995
Classification name: PACS per 21 CFR 892.2050
Regulatory Class: II
Product Code: LLZ

Model: AquariusNET Server
Manufacturer: TeraRecon, Inc.
510 (k): K012086
Classification name: PACS per 21 CFR 892.2050
System, Image Processing per CFR 892.2020
Regulatory Class: II
Product Code: 90-LLZ, 90-LMD

And that's what GE said about the AW Server! In this section of this letter, the Indications for Use are as follows:

AW Server is a medical software system that allows multiple users to remotely access AW applications from compatible computers on a network. The system allows networking, selection, processing and filming of multimodality DICOM images.

Both the client and server software are only for use with off the shelf hardware technology that meets defined minimum specifications.

The device is not intended for diagnosis of mammography images. The device is not intended for diagnosis of lossy compressed images. For other images, rained physicians ay use the images as a basis for diagnosis upon ensuring that monitor quality, ambient light conditions and image compression ratios are consistent with clinical application.

Remember, GE wrote these specs, and the government approved them. Reading this paragraph makes the question of what is allowed for diagnosis and what is not even less obvious.

I understand where GE is coming from with their claim, and I like the "Smart Compression" approach. Still, the claim to be the "only FDA approved remote review system" is very bold. If true, and if it stays true, I'll very strongly consider purchasing the AW Server, if and when I get the cash to do so.

Again, I invite the commentary of the other major players in this space. What say you?

4 comments :

Anonymous said...

This kind of confusing self definition non-sense (which occurs for all companies in this domain space) would easily be resolved if we had clearer standards for what is defined as a diagnostic quality image. This would differ for each modality, but its critical moving forwards in the realm of remote viewing for the standards bodies to establish these minimum standards.

In the cardiology world which I work in, there are only very lose interpretations which govern the minimum leaving vendors pointing fingers at each other especially for things like remote viewing. What CANNOT happen however is such a loose interpretation that would allow for bad clinical practice where the end user would be compromised by overly compressing. Ultimately the customer loses.

Anonymous said...

My experience is once pushed onto PACS all beats are off.

Anonymous said...

Did you know that Carestream (formerly aka Kodak Health) already had a native full DICOM streaming solution since 2007 ?

With adaptive intelligent scrolling, multilayers, and optimized DICOM communication. You start reading the first images lossy, and in seconds this automaticly shift to lossless !

On top of that you have all the common post-processing tools you need AND native (!), from MPR,VR to fusion, multimodality volumetric matching and even PET/CT toolset.

All the things you highlight in your latest post as "WIP" from GE, Siemens etc...already available for 2 years...and FDA cleared...

Anyway. Beats even Amicas ;-)

JG said...

I wanted to note that the Terarecon APS product cited is not the image rendering/display part of the architecture. It is an add-on server that enables user-defined pre-processing. For example, you can route all CTA Abdomens through APS and autoremove bone in advance of viewing. The part of the system that would be relevant would be AquariusNET or also the more enhanced Aquarius Intuition product.