Tuesday, December 31, 2013

RSNA 2013: My top 10 on what’s new and what’s old, part 2.

It is the second day of the RSNA radiology tradeshow in Chicago. I am flagging down a cab to get me to my 7 am breakfast meeting in the Hilton Towers, which is on my way to the conference at the McCormick Place. After this meeting, we hurry into another cab to get us to our courses, lectures, and appointments at Starbucks. I always have my phone close by to look for texts from those I am going to meet, who might be stuck waiting for the bus, held up with another appointment, or were called into a last minute customer meeting that has a higher priority than meeting with yet another consultant like myself.

Walking the floors...
Meeting with many professionals and walking the floors is how I find out what’s new. After having commented on what I saw that was truly new, gadgets and/or product and services in part 1 of this series, here are my observations on what is not really new, which means that I have seen it in previous year(s), but it has reached either a new level of maturity, or added significant improvements or new features this time around. I therefore label these as “new and old.” I’ll address any noteworthy “truly old” developments in part 3. So, here is my next top ten (or maybe 20) listed:

 your workflow looks less complex
      1.       Workflow: Workflow is an “old” subject. Vendors have been talking about it for many years, however, it still seems to be a major struggle. When doing my informal survey of the top three issues facing customers and vendors, workflow is almost always the number one issue. One should think that after 20 plus years of PACS implementations, it should have been addressed and solved, however, nothing is further from the truth. Why are we still struggling with this subject, and why is it that according to many of my colleague consultants, 60 percent of the hospitals run at a sub-optimal level and could achieve major improvements in efficiency if they would take the time to look at what they have and what they could have? In my opinion, there are four reasons they don’t do this:

·         Institutions and staff don’t take the time to do long term planning. Every PACS administrator I’ve talked with is in the middle of a PACS upgrade, or having trouble keeping up with all the changes required for Meaningful Use, or is just busy fixing studies and addressing other burning issues. Note that this is only from the ones I talked with who were sent to the RSNA, not those who did not have time or could not get the funding to travel to this event. No one seems to take time out to sit down with all parties involved and see how the workflow can be improved.
·         There is a big lack of knowledge of the clinical workflow among the people managing these systems. PACS systems were initially managed by radiology staff, with a PACS administrator reporting to the radiology administrator, who properly can set priorities and understand the workflow implications of changes, upgrades and new interfaces. PACSs are no longer just a radiology project but have become an enterprise activity taking care of managing images from multiple departments and specialties, and have migrated to IT for support of the hardware, (which has been located to their central computing center), and software (helpdesk support etc.). The most common complaint I hear is that access to the servers is now limited and a simple reboot or just checking status and/or files on the servers is close to impossible as they are locked out of accessing the main computer center, and that any support call now has to go through another two layers before it can get to the person involved. Needless to say knowledge of clinical workflows is greatly watered down at the support level with people who only have a pure IT background taking care of business.
·         IHE recommendations and profile definitions are still being ignored. One of the first profiles defined by the Integrating the Healthcare Enterprise (IHE) is called “scheduled workflow.” It was subsequently followed by an “unscheduled workflow” profile definition addressing the case when incomplete patient data is available for trauma cases. Standard DICOM and HL7 transactions were refined, and options eliminated as much as possible to come to a rigid and well-defined sequence of transactions. This results in the automation of changes in procedures, patient updates, and automatic verification by the modality of a study to allow for modality and workstation worklist synchronization. However, when I ask my students in our PACS training classes how many have implemented Modality Performed Procedure Step and Storage Commitment, which are essential components of the scheduled workflow profile and used to streamline the department workflow, I get either blank stares or at best a confirmation of maybe 10 percent of the audience that they have implemented this. One of the reasons goes back to my first comment, i.e. people being too busy with day-to-day activities to take the time to sit back and look at how to fundamentally change their workflow to really make use the technology they have available.
·         It is not about just the PACS anymore. Assuming that an institution has a well functioning PACS system, there are several other subsystems that have to work correctly, such as the critical results reporting and ER discrepancy reporting. In addition, to ensuring proper quality of care, one should also address the peer review process. With regard to the critical results reporting, there is nothing more frustrating to a radiologist than having to track down a physician to follow up on a critical finding marked code red, if there has just been a shift change, or when the finding occurs at the end of the day, or when the patient already left the ER having been sent home by the physician who missed the critical issue. Increasing integration with multiple systems is lacking (see integration comments below as well).

There are still vendors
who claim that their archive is a VNA,
despite the fact that they merely have
a simple DICOM Store (level 1)
2.       VNA: The implementation of VNA’s has moved from what the Gartner consulting group calls the Technology Trigger, passing through the Peak of Inflated Expectations to the Trough of disillusionment phase. In layman’s terms, it did not meet the expectations of the initial hype as people found that a VNA implementation has some major challenges.

I found three types of VNA vendors, the first group are those who embrace it with both arms as they see that a full level 5 implementation (see related white paper of the different levels) gives them a strategic advantage and they seem to be very well positioned to address customer needs.

The second group are the laggards who are trying to catch up with adding the functions needed for

The third group are those who interestingly enough are still ignoring the need to offer a full-fledged VNA and think they can just put another label on their existing PACS archive and hope that clients are not looking through the marketing smoke and mirrors and recognize that this is just the same old thing. One vender told me jokingly, upon asking me what I could do for him, that he wished I could take the VNA away, but unfortunately I don’t have a magic wand, and even if I had, I would not want to use it to turn the clock back 10 years, as VNA’s are here to stay and addressing the issue of providing a true enterprise image information and management solution using open standards is essential.
a full VNA, such as synchronization between multiple PACS systems and VNA, full-featured routing and pre-fetching, tag morphing, information life-cycle management and support for a uni-viewer and HIE connectivity. These vendors have obviously underestimated the demand and are working hard to catch up. Many of them are learning the hard way by deploying VNA’s prematurely resulting in all types of workflow and other issues.

HIE's look great on paper but are still
far from 100% operational
3.       HIE and XDS: The establishment of Health Information Exchanges (HIE’s), either private or public, have been a major part of the US government’s initiatives to facilitate information exchanges to reduce unnecessary duplication of tests, and making observations and results widely available to health care practitioners. A key requirement of the HIE is the support for the Cross Document image and information exchange standards, aka XDS. XDS is not rocket science, it is based on existing standards, however, there are very few implementations of it. The reason for the relatively few implementations appears to be (according to the people I spoke with during this event) a gross misunderstanding of how it works, what it does, and what is needed for existing infrastructure to support it. Hopefully, a better understanding by training and education by the IHE committee and other third parties (here is a shameless plug for the training provided by OTech) will change this.

"big-foot" vs "zero-footprint"
       4.       Uni-viewer: Last year’s zero-footprint viewer is re-labeled as a uni-viewer. However, I would expect that a uni-viewer does not only have the typical zero-footprint characteristics such as leaving “no-trace behind” after a physician logs off, but also can display other specialties such as dentistry, ophthalmology and obviously cardiology and in the future pathology. I have seen one uni-viewer that also supports the new “enhanced” CT, MR, cardiology, and angiography image specifications (aka DICOM SOP Class support) and even digital mammography multi-slice tomosynthesis objects. For these large objects, server-side rendering seems to be the best solution as transferring one of those studies onto local cache will take too much time. Many of these viewers are positioned to connect to a VNA, using, for example, the new DICOM web-based protocols, and provide patient-centric vs a department-centric approach to physicians. Interestingly enough, I talked with one institution where the radiologists have taken notice of the fact that physicians have ready access to patient studies from multiple locations through this uni-viewer while accessing the VNA and they also want the same functionality. They don’t quite realize that the PACS has a rather sophisticated workflow manager functionality that interfaces with their radiology viewers and provides a worklist that is synchronized between multiple readers and maintains the status of when studies are read and reported. However, as one person commented, as soon as VNA is able to provide this functionality, the “PACS might be dead.”

5.       System integration: The first radiologist I ran into at the RSNA meeting mentioned that poor
The more "bubbles" the more complex
integration is his main issue. He said he is reading images from multiple facilities using a so-called worklist aggregator, which communicates between the several PACS vendors and provides a composite worklist to his workstation. This software needs to integrate with his PACS systems, his voice recognition system and the RIS. In addition, the system interfaces with the EMR to export the reports as well as images. Therefore, in total he has four systems that need to be integrated. It used to be two (PACS and RIS), followed by the voice recognition and now the EMR as well. It appears that the level of integration and associated complexity has reached the upper limit as he is struggling with several interface and integration issues. Just coordinating the synchronization of these systems with regards to upgrades is likely a major issue, for example, if the PACS is upgraded, it has to be tested and verified with four other systems. Imagine if he had an additional peer review system, and critical results reporting system, as well as dose reporting system. I don’t know if that is going to be manageable; it seems that the limit for systems integration is about to be reached.

Note integrated phone and
capability to access images
from the bed side monitor
6.       Bedside integration: Anyone who has visited a hospital room lately has seen the COW’s (Computer on wheels), that are used by the nursing staff to take vitals and other information about a patient. Increasingly, the clinical features are integrated with patient resources and entertainment at the bedside through a stationary monitor. One vendor demonstrated this whereby this terminal can show not only your movie on demand, but also pull up a patient record from an EMR, including any corresponding images, take vitals, and even allow for a teleconference with a physician or nurse using the small camera, (which can be physically covered when privacy is needed). A wireless keyboard is provided as well.
Gesture controls for view stations
(note game controller on top of monitor)

       7.      The use of gaming controls to view images: Last year I saw an adaptation of one of the gaming consoles to use gestures to control a monitor. This year there were several demonstrations, both as part of the scientific exhibits and built into commercial products. This application is primarily for use in conference rooms and for teaching whereby a physician can remotely manipulate images.

8.       Auto-scrolling through image stacks: Gesture controls will not help much for day-to-day use by
Note small rectangular autoscroll
hardware device in between mouse
and computer
radiologists in their regular reading. However, a new auto-scroll device might help them as many are suffering from wrist issues from using the mouse and/or trackball day in and day out. This device, which simply connects between the trackball and the computer will automatically scroll the images, which is especially important for axial images sets such as for CT and MRI. Hopefully, this will prevent some carpal-tunnel syndrome sufferers in the future.

Demonstration of a
surgery procedure
        9.       Virtual anatomy tables: also this year there were multiple vendors showing, what is known as the “virtual pathology table,” which is basically a large touch screen display laying flat and built into a demonstration table which allows for a physician to manipulate data that is typically based on a CT ort MRI 3D data set thereby performing a “virtual autopsy.” This is a great teaching tool, and can also be used for forensic applications in case a person has been buried provided they had a CT scan done in case there are follow-up investigations.

        10.   Dose: The registration of radiation dose has become even more important as several states have enacted legislation or have bills pending approval that require dose registration of CT scans as a minimum, and potentially other X-ray exams in the future. Every institution in the affected states is scrambling to establish a set of policies and procedures matching the technical capabilities that are available with additional software solutions. Unfortunately, existing systems have to be upgraded to export the well-defined DICOM structured reports that contain the dose information, which means that in the meantime, several vendors have implemented band-aid solutions, which rely on screen-saved dose overview information, which has to be interpreted using OCR (Optical Character Recognition) software. There seems to be a consensus that this information is to be stored at the patient level for example, with his or her personal health record, however that infrastructure is not (yet) in place and therefore many store the information in a EMR, or even a PACS or RIS where it obviously does not belong. It will take a few more years for dose registration to become ubiquitous and be seamlessly integrated into the regular workflow.

In conclusion, the most heard “old topics” were VNA and uni-viewers, workflow, HIE and XDS. These are still immature and will need several years to come to fruition, I am sure we will see them again at next year’s tradeshow. In the mean time, look for part 3 of this series coming up soon.