Friday, July 20, 2012

Office imaging systems revisited

Summary: An office-based CD importation system with a single centralized store and viewer dramatically improves ambulatory workflow efficiency (as one would expect).

Long Version.

An excellent article this week in JACR, entitled "Streamlining the Receipt and Upload of Diagnostic Images in a Clinic Setting", by Bartels et al from the UCLA Dept. of Neurosurgery, reminded me that it was time to take another look at the matter of how clinicians are coping with images, since it has been over four years (!) since I wrote a post about "Requirements for an 'office imaging system' for requesting practitioners".

The UCLA article addresses the entire process of receiving outside images through to viewing by the neurosurgeon, relatively formally as a process improvement initiative using the Lean (Toyota) methodology. The net result is summarized as reducing the neurosurgeon involvement in CD handling from 91% prior to the improvement to 5% after, largely as a result of introducing an automated import process executed by the clinic staff and providing a standard viewer for pre-loaded images. Automation of the staff import process was also described as significantly reducing their time.

The process is described as opening the (DICOM) CD (in iQ-View perhaps, or possibly in some GQC-specific importer), pushing to the Global Care Quest (GCQ) Neurosurgery cloud-based PACS, and using some pre-caching (auto-load) strategy that was then improved by the use of higher bandwidth connections (from 10Mb/s to 1 Gb/s). The process handles CDs brought prior to the visit as well as those brought to the clinic "at the last minute".

Non-DICOM CDs are specifically mentioned as being handled with a "red sticker" to cue the neurosurgeon to fire up the native CD viewer; how often these are encountered is not mentioned explicitly, but presumably they account for much if not all of the residual "5%" that requires physician handling the CD.

It is gratifying to see that this relatively simple process is indeed feasible and practical in an ambulatory setting, and furthermore on a relatively large scale, using ordinary components, despite some comments from an orthopedic surgeon on my previous blog entry that this sort if thing was "completely unrealistic".

We have certainly seen over the last decade that importation of outside images (and reports) into the hospital (enterprise) PACS has become routine and the standard of care, but it is gratifying to see that the ambulatory setting is being addressed too.

The article was a bit light on technical details though, particularly of the existing infrastructure that was being re-used, and I would be interested to know why they use a dedicated neurosurgery PACS, and also why it is cloud-based. Digging into this a bit deeper, it turns out that Global Care Quest (GCQ) is actually a commercial spin-off from this same UCLA neurosurgery group, and this description in the press describes some of the disparate data source, multiple site and mobile device integration problems that it was designed to address. There is a video of one of the authors, Neil Martin, who is the chairman of Neurosurgery, describing it. I get the impression that since they have sophisticated critical care informatics solution in place, it made sense to leverage it for the ambulatory setting too, though that does raise the question of how smaller less sophisticated "offices" would address the same problem without a pre-existing infrastructure in place.

Locally obtained images were described in the article as being pre-fetched into the neurosurgery PACS from the UCLA PACS, so that a single viewer provided a longitudinal view.

One has to wonder what all the other ambulatory departments that are heavy image users (like orthopedics, neurology, rheumatology, pulmonology, etc.) are doing about importing and why there isn't a pan-UCLA solution in place.

Certainly, using an outside provider to host the system in which the imported images live makes a lot of sense, compared to neurosurgery having its own IT department (i.e., the usual "software as a service" benefits), but given the large size of images and the cost of bandwidth (which the authors noted as an issue that they addressed), having an on-site archive (even if managed by outsiders as a contracted service) might make more sense in other situations. I.e., "outsourcing" does not necessarily imply the need for "off-site". For smaller offices for whom external access or collaboration is not a requirement, for whom long term archival is not an issue, and who cannot afford the bandwidth, on-site may be a better option.

The article also did not specifically address the matter of identifier reconciliation on import, and whether this is handled at all, manually, or using an automated approach (whether based on IHE Import Reconciliation Workflow (IRWF) or homegrown procedures).

In the long term, of course, we all hope that CDs (or any other form of physical media) will go away, as the producers of images find more "modern" ways to get them to the consumers, whether it be via external portals indexed via registries (a la the original XDS-I model), or regional (central) repositories (a la the Canadian DI-r or various "cloud-based" image storage and distribution providers), informal push networks, or just by inclusion of images as part of every patients' external PHR, such as espoused by the RSNA's Image Share project. Until questions such as those about cost, scalability, sustainability and access control are addressed though, for many people these more ambitious approaches remain elusive.

But in the short term, the UCLA Neurosurgery experience is further evidence that an individual ambulatory facility can take control of their own destiny, minimize the inconvenience and mazimize the utility of the existing CD-based infrastructure, using tools that are becoming, if not already, off-the-shelf commodities.


PS. There do now exist third-party importers that claim to address some if not all the proprietary formats, including Amicas and iSite (iSyntax), so even the residual "5%" problem could be addressed with a more sophisticated importer, though admittedly there will always remain the problem of unburned, un-finalized or physically damaged CDs. I did take a crack at the Amicas CDs myself a while back (see "ConvertAmicasJPEG2000FilesetToDicom"), but reverse engineering the iSyntax format looked like way too much effort, so I am glad somebody has implemented it.

PPS. The article does mention training the surgeons to use a single "universal" viewer; I wasn't able to tell from the GCQ site whether or not their viewer was compliant with the IHE Basic Image Review (BIR) profile, which was specifically developed with neurosurgeon involvement to satisfy their review needs; as far as I know, GCQ has not tested the product at an IHE Connectathon.

PPPS. After the recent debate in the UK Imaging Informatics Group, I hardly dare mention reports, and it may be that the neurosurgeons are not all that interested in the radiologists' reports, but I would like to know how and if the outside reports are being imported if present on the CD or provided in some other manner (fax, email, scanned paper document, EHR, PHR, RIS, HL7 V2 feed) and how they are handled in GCQ and linked to the images.

PPPPS. There has a lot written about the mechanics, benefits and policy issues of importation of CDs before, and I wouldn't want to ignore the literature on this subject, a lot of which can be found with a Google Scholar search, a search of the Aunt Minnie Forums, and Yahoo pacsadmin, as well as various SIIM materials; I call attention to this article though, because it is specific to an ambulatory setting and involves non-radiologists. I was a bit surprised though, when I realized that the article had absolutely no references at all (and that this got past the reviewers and editors)!