Friday, January 30, 2015

Top Ten PACS Support Issues.

PACS administrators are always busy fixing problems, however, there should be a trade-off between
fixes, planning and other support activities.  Talking with seasoned PACS administrators, they believe in the 50/30/20 rule: you should spend 50 percent of your time fixing problems, 30 percent on training, creating and checking on policies and procedures, project management, and 20 percent on new projects. If you are spending more than 50 percent on fixing problems, you have a major issue, i.e. there are too few resources, the PACS system is unreliable, there are serious workflow problems, technologists and/or data input personnel are not taking responsibility for their actions, or there are other fundamental issues.

Here are the top ten issues that PACS professionals are mostly concerned with when spending 50 percent of their time fixing problems:

1.       Fix broken, or unverified studies – This seems to be the overriding issue. As the PACS administrator is ultimately responsible for the data integrity of the PACS system, he or she is the go-to person if a study does not make it to the worklist of the radiologist. A study ends up in the “penalty box” of the PACS system because information is missing from the metadata, i.e. DICOM header information such as patient demographics, identifiers, Accession Numbers, or there are duplicates and/or conflicts (such as another patient using the same ID). The good news is that PACS vendors’ tools have gotten better, making it easier to fix these studies, which in many institutions allows for a “super-tech” or PACS assistant to take over these chores from the PACS administrator.
2.       Modality Worklist issues – Interestingly enough, even though the worklist is a very mature DICOM protocol and service, I still hear about issues with these worklists. It seems to be related to the fact that now almost every PACS or RIS creates its own worklist, which is often less mature than when everyone was using the de-facto standard Mitra broker. Also, many new modalities such as used in surgery (endoscopy), cardiology (intra-vascular OCT-IVOCT), dentistry (cone beam CT-CBCT), and ophthalmology are now being connected to the PACS, and they are often not as mature with regard to their worklist implementations. The modality worklist has to map information for the order such as, patient location, status, and procedure codes, to the modality and station name and/or Station AE title, which often is not done quite right. The result might be that all of the radiology procedures end up at the worklist for the Dexa bone scanner as the worklist can only filter on modality being CR, for example, or that all of the in- and out-patients appear on the list of the outpatient clinic at the CT, and other mismatches.
3.       CD import issues – Many issues are still occurring with images being brought in from the outside, e.g. on a CD. Issues range from having proprietary compressed images not DICOM images, to a, missing directory on the CD, or encoding issues in the header that the PACS might complain about, which needs manual fixing. The general rule of every PACS system should be, to be as conservative as possible when creating images, especially when they are to be sent out, i.e. create plain vanilla encoded information that follows the DICOM standard, and be as liberal as possible when reading in images, i.e. be prepared, many vendors make mistakes. That is not to say that one should blindly import these rogue images, but, rather, be able to view them and fix them in case they are to be archived locally. An example that I encountered with a CD is when I tried to display the X-ray of my dog that the veterinarian gave me and the images did not open up in my viewer with the message “invalid header.” It appeared that the patient ID tag was missing, obviously a gross error, but no reason to not display it as I had to use a DICOM editor to look at the header and fix it before I could review them. When the DICOM violation is not obvious to the naked eye, I run a validator which tests the header against the data dictionary and provides the information that I need to be able to fix it.
4.       Protocol, i.e. negotiation problems – These problems are caused by non-SOP Class support, for example, when a new breast tomosynthesis device tries to send its mammography slices to the PACS, or when a dose structured report from a CT is being rejected. Another cause could be non-support for specific compression schemes such as wavelet (JPEG2000), MPEG4, or other advanced compression schemes. It is critical to consult the appropriate documentation, i.e. DICOM conformance statements, and, if the issue is not obvious, one can use a DICOM sniffer to see exactly what is proposed by the DICOM sender and what the response is back from the receiver. Other protocol issues might occur when there is a different behavior expected by the receiver than by the sender, for example, if using a Storage Commitment, the receiver expects the response to send back a different Association while the sender is waiting on the same Association for the result to come back.
5.       Connectivity issues – These are mostly network related, i.e. a lease for an assigned IP address might have expired, a network administrator might expect every device to run DHCP and handling dynamic addresses, ports being shut down, routers having issues with auto-negotiating the full-duplex mode with some older modalities, etc. Performance issues and security threats are the most common with wireless connections, for example, for the portable X-ray units in the ER or on the floors.
6.       Image quality issues – There could be windowing and LUT support issues, not necessarily with the more mature modalities, but mostly with PET/CT images, for example, or support for Presentation States might be missing. Then there is also obviously the monitor calibration issue. After all these years, there are still many users that don’t realize that viewing a chest CR or DR image on a commercial grade, uncalibrated monitor is not only a major disservice to the patient and the resulting diagnosis, but sets you up for a malpractice lawsuit in case a missed finding causes major harm. One of my PACS students actually told me that he had to pull out the calibration records of the PACS monitors for several years back as they was used as evidence in a court case.
7.       Hanging or display (DDP) protocol issues – When images appear on a workstation upside down, in reverse order, left/right mixed up, or in a different series requiring yet another mouse click by a radiologist, that means that the software is unable to make sense of the information in the DICOM header to make sure that they are displayed in the correct order and position on several screens. Reasons range from missing or inconsistent body part descriptions, protocol definitions, or software that is not intelligent enough to look at orientation and spacing information in the header but relies on image or series numbers. This problem is getting worse as images are now increasingly being exchanged between multiple facilities, which means that even if your institution has well defined protocols, and series and study descriptions which are consistently used, they might differ from what is used in another facility. Note that these inconsistencies do not only impact the display but also the capability to pre-fetch prior exams.
8.       Random errors – Random errors include occasional dropped connections, missing images, the system going down for whatever reason, unexpected aborts or resets, etc. These are hard to diagnose and/or solve. If the problem is semi-random, for example, when sending images from one CT, on average one out of 10 studies is received incomplete, one could attach a network sniffer, set it up to capture the communication in a rotating buffer and review that after an error occurred. These errors also greatly depend on the hardware used and the vendor. I have talked with PACS administrators who use a UNIX backend and have not rebooted their system in two years, but I also talked with one who has a system that goes down every week (and this is from a major, well known vendor!)
9.       Workflow issues – When talking with radiologists about issues, their number one issue is always “workflow,” which means that the current PACS systems do not quite meet the need of their day-to-day work in a productive and efficient manner. For example, a radiologist might have to log into three different accounts when reading from three hospitals, instead of having a single worklist, or technologists need to do manual checks and balances to complete studies in the RIS and/or PACS. Another workflow issue stems from procedures changes not being automatically communicated by services such as the Modality Performed Procedure Step (MPPS) that require manual fixing, and many others. Workflow issues seem to have gotten worse with the many poor Vendor Neutral Archive (VNA) implementations requiring the fixing of studies in both the PACS and the VNA instead of having them synchronize the changes using the appropriate standard implementations. The frustrating aspect of this is that there are technical solutions, most of them specified in detail by IHE to have an automated workflow, and there are third party general purpose worklist aggregators that can be used, but in most cases they are either not provided (yet) by the vendors, or if they are, they are not used due to a lack of knowledge by the local users. A comprehensive workflow analysis identifying how the existing technology and proper workarounds can assist in becoming more effective and efficient would go a long way to resolving some of these workflow issues. A PACS administrator is often in a great position to do this.
10.   Missing support for important features – Upgrading a PACS system to a new release is often associated with a substantial additional cost, effort, and risk if it impacts the current workflow and efficiency. It would be interesting to do a poll on how many users are up to the latest release, I bet that it is probably less than 20 percent. That means that, for example, support for dose structured reports, measurements from ultrasound, new modality support such as breast tomosynthesis, is often not available. That means workarounds, less optimal solutions, and often additional work for the PACS administrator. As long as budgets are tight, this is not likely to change.

In summary, there are many issues that that a PACS administrator faces, but again if this takes more than half of your time, there is less time for proper planning, training, and preventive activities, which will result in a vicious circle of always trying to catch up by constantly fighting fires. If you are in that position, you might consider a serious talk with your supervisor. Good luck!