Monday, November 12, 2012

How to deal with finger-pointing between Imaging vendors.

As healthcare imaging and IT systems are getting more complex and the number of systems to be integrated is increasing, it gets harder to identify and troubleshoot interoperability issues. Information crosses several systems boundaries, several of which are not under the control of a healthcare imaging and IT professional, and almost always from different vendors. Upgrades and changes can occur at various systems and subsystems adversely impacting the operation. 

The key to resolving these issues is first of all locating the area of concern and then second visualizing it. This will assist the vendors who are involved to address the issue without them finger-pointing to each and not taking any immediate action.

An image might be incorrectly identified and/or processed wrong because of incorrect header information, which might have been initiated by an error in the personal Health Record (PHR), which was loaded into a Centralized Physician Ordering (CPOE) system, which placed an order through an interface engine, onto a RIS scheduler to a modality worklist broker, which was queried at a modality, copied the information in the image header, archived at the PACS, and retrieved at a viewer plug-in showing this information to a physician looking at an electronic health record of this patient.

To be able to address these kinds of issues, a Healthcare Imaging and IT professional needs to follow a systematic approach for locating the cause of the problem using a decision tree and then using the appropriate tools to visualize the issue. The good news is that vendors have stepped up over the past few years to increase their capability for logging, auditing and monitoring their interfaces, however, in many cases the errors are still vague and not to the point. Examples of such vague errors are “processing errors”, time-outs due to unidentified problems, resets or aborts, and many others.

In addition, there are also more tools available, most of them in the public domain that visualize issues at many levels of the interfaces, as detailed as the actual bits and bytes that are exchanged between the devices. The only barriers to using these tools is a general lack of knowledge and training of healthcare imaging and IT professionals as well as the vendor service personnel, and in some cases the lack of access to networks and routers due to security concerns by IT departments. The latter can in many cases be resolved by partnering with the people who are responsible for the IT infrastructure and try to get them involved with the resolution of the issue.

The first step in diagnosing the interoperability issue is to characterize and identify the type of issue. Tools to perform the diagnosis can be grouped as follows:
·         Utilities, such as accessible through a command line interface or service menu. These can be used to test basic connectivity for example by a ping or DICOM Echo.
·         Active simulators such as modality worklist simulators (see link for demo), RIS/PACS simulators, and viewers, all of them are available in the public domain.
·         Passive tools such as DICOM sniffers (see link for demo), also available for free which can not only make the information exchange visible but allows saving these interactions and have them processed by Validators to find out if there are any violations and/or issues with the data formats or protocol.
·         Validators to validate data formats (headers) as well as the protocol (see link for demo). Fortunately vendors who have developed an extensive set of validators also have made these libraries and utilities available in the public domain.
·         Test transactions in the form of scripts, and many test images to evaluate image quality as well as the image processing pipeline are available, mostly as a byproduct from the many IHE connectathons.

The interoperability issues that are to be identified using the tools above can be categorized into four areas:
·         Connectivity errors, which can be due to networking issues, incorrect addressing, problems with negotiating a connection between the applications, performance issues and status errors.
·         Display errors can be related to worklist issues for example, populating a worklist incorrectly. The correct display of the image and related information, hanging protocols, incorrect handling of Structured Reports, such as used to display measurements, CAD marks, identify Key images, or other information such as radiation dose. Overlays and presentations state information is a category by itself, including on how to handle incorrect “burned-in” text.
·         Image quality issues can be hard to identify as the source can be the image acquisition, modality processing, view station imaging pipeline or display itself. Test images and test objects inserted at various locations in the imaging chain will assist in troubleshooting these.
·         Exchange media problems are getting less but still present due to non-DICOM compliant CD’s being created that might have non-DICOM images, lacking a DICOMDIR, or stored in format not supported by the particular DICOM profile definition.

In conclusion, in order to troubleshoot interoperability issues, the first step is to follow a decision tree in identifying the type of problem, than selecting the appropriate tools to visualize it. Despite increasing complexity and many additional systems that are to be integrated, the availability of tools in the public domain, makes troubleshooting and diagnosing problems possible to be performed by Healthcare Imaging and IT professionals.