Saturday, May 07, 2005


One of the sites we cover signed up for WarpSpeed from Image Technology Laboratories. After meeting with just about their entire cadre of execs and engineers, here are my thoughts (sent by letter to the owner of the site in question):

I would like to thank you for yesterday’s opportunity to visit with the Image Technology Laboratories team. The meeting was quite interesting, and their engineers are obviously quite knowledgeable. I would go so far as to say that were I independently wealthy, with a lot of extra time on my hands, I would buy the company myself and with the expertise available, create a super product.

I promised you an objective assessment of the WarpSpeed product, and I will do my best to provide this. In brief, the software that is most important to me, the radiologists’ reading client or viewer, is not ready for deployment. In all honesty, based on our discussions yesterday, I would place their product about two to three years behind industry standard. While they have many good (even superb) ideas, ITL has not yet assembled them into a system that I would be comfortable using in day to day practice. There are a number of factors that lead me to this conclusion, and I would be glad to discuss them with you at length if you really want to take the time to do so. Let me cite one example which really tells the story. You might recall my question about linkage of two CT studies, and whether they use image number or table position. They really seemed to have no idea about the difference between the two approaches. The use of table position linkage allows the comparison of scans performed at differing slice thicknesses, which is a fairly common situation. Linkage by table position has been industry standard for at least the past 2-3 years. I think this is really telling of the entire direction of ITL. They are an assembly of very sharp and high-level engineers, but I am assuming that the radiologists with whom they partnered had little PACS experience. The entire approach ITL is using with their product shows innovation, but also near-complete isolation from competing approaches. Their data distribution structure is reminiscent of Agfa and Siemens designs from five to eight years ago. The majority of systems today are web-based, relying on the architecture of the Internet itself, rather than reinventing the wheel. Their concept of a thick-client viewer is still used by some systems, although many today use web-based clients or viewers. Placement of a computer between each modality and the main server is an approach very rarely seen today. Finally, the method that was outlined for remote access is close to unworkable. It attempts to isolate the user’s computer to “prevent transmission of viruses” and does so with a great deal of effort. Most systems today use the web approach for remote access, which has proven to be safe if implemented properly, and requires no significant modification of the remote computer.

From a hardware standpoint, I would be very concerned about their proprietary assembly of the core of the system. The fact that they use “Intel motherboards” does not assuage this worry. The majority of vendors have gone to using “off the shelf” servers, storage, and the like, lowering the cost and allowing easier service. A proprietary collection of components (even if those components themselves are common) could present a service nightmare.

I do not claim to have the business expertise you possess, but a quick review of ITL at raises several further concerns. According to SEC filings, they had a significant net loss last year in spite of increased revenue. The contract with you is very prominently mentioned, and it is rather obvious that there have been no other recent sales. This is a company operating at the very edge. Their future existence is really in question. They will depend on sales to sites with no PACS expertise, and those are becoming fewer and further in between as the technology continues to grow.

What concerns me the most with this company is that their product is for all intent and purpose in beta. It is NOT ready for a group with our level of PACS experience. To get them there will require an amount of testing, time, and patience that we simply do not have. Your operation is very busy, and your scanners churn out a rather significant number of images. We cannot just stop reading those studies to call up the folks in Kinsgston and work out
problems. We do not have the time (or in the case of most of my partners, the expertise) to develop this product. We have been down this road before with ScImage. When my former partner first brought it to us, it was buggy, it crashed constantly, and it was very painful to actually use. This was a company that should have gone out of business. My partner was enamoured with their 3D module, and was willing to ignore the poor programming of the entire remainder of the system. Rumor has it that he invested a substantial amount of his own money to keep ScImage afloat, and they are still around today with a more stable (though still poorly interfaced in my opinion) product. You may have seen our new computer in the reading room which we use for the orthopedic studies. This utilizes the StorComm MedView client, which is again very tedious and completely unintuitive. I am fielding about 4-5 calls per day from our radiologists struggling with this less-than-optimal software.

With the above in mind, I must urge you in the strongest possible terms to reverse your purchase of the ITL system. There are other alternatives which I would be more than happy to help you pursue. Please contact me at your earliest convenience so we can discuss this situation further.

No comments :