Monday, November 15, 2010

Enterprise Image Management

I've written about our multi-year effort to archive all types of medical images in a single storage platform across the enterprise, making every image viewable everywhere to authorized clinicians.   Here's an overview of my lessons learned thus far on the journey.

Image systems have been notorious for being isolated pockets of technology.  They are often supported by mini-IS sections at the department level.  Vendor support requirements have been significant, often requiring a dedicated on-site engineer at larger institutions.  Infrastructure components such as storage, network switches, and workstations have often been only available through the image application vendor.  Configurations have often been proprietary and  have not lent themselves to substitution by technologies of the customer's choice.

There has been rapid transition from analog to digital imaging, advancements in imaging (32 and 64 slice), and growth in procedure volumes.   The desire to archive digital images for medical, research and legal reasons has increased. The above has made image storage a costly line item in capital budgets.

Thus, rather than continue managing imaging systems as isolated environments, our goal is to create a common storage archive and I've achieved unified departmental support.

Attributes of the archive include
*Low cost - a total cost of ownership less than $1/raw GB initially and less than $.50/raw GB within four years.
*Response times that exceed those of departmental archives in use today.
*Support for both DICOM and non-DICOM images.
*Support for life cycle image management (purge, compress, tier migration)
*High availability and disaster recovery protection
*Multi-protocol support (CIFS, NFS, REST)
*Central IT management and budgeting.

There are many emerging products from companies such as Teramedica, CareStreanm GE, EMC, and Dell.   There are cloud services available from Symantec, Life Images, and Accelarad. The past few years have seen a gradual migration of image management backends from departments to central IS organizations.    In the next few years, I expect departmental experts to focus on imaging modalities and specialized front end applications leaving servers and storage systems to be managed by IS.


Joe Stewart said...

Perhaps it's obvious, but just to be sure, I'd include the following essential attributes:

1. access to content must be trivial for the clinician and in context. The easiest technique is to insert links to content directly in the patient's record thus providing the clinician one UI from which to access all clinically relevant information. Furrthermore, those links need to be organized in such a way as to be completely obvious to the clinician - perhaps on an "imaging" tab or from within each relevant order or encounter.

2. both diagnostic and "review" quality displays of the content need to be available. Diagnostic viewing is commonly performed at calibrated workstations while "review" viewing is done with patients wherever it is convenient.

Stephane Urlot said...

Are you considering a move to public clouds or building a private cloud ?

Miguel Cabrer said...

This is really challenging and has different approaches. From an Enterprise can decide to store all non radiology images (dermatology, ophthalmology, path slides, ECG's) into the PACS as dicom objects. That's a "Beyond PACS" approach. Another approach is to keep the PACS as Radiology system only and use a MMA (Multimedia Medical Archive) for managing all rest of media including video.

In any case, when reviewing we can find some other scenarios (multi PACS organisation, different vendors, repositories, etc.). Including an URL in the Medical Record can be a solution but also a multi-speciality light viewer (web embeded) combined with WADO is much elegant solution.

I am happy to hear this is becoming a need worldwide. We have good experience on that field.