Friday, June 5, 2009

Cool Technology of the Week

Today's Cool Technology blog entry is not about a product, but a concept.

I've had numerous companies (more than 5) approach me in the last 90 days with a product in development that I'll call "Image Exchange in the Cloud".

One of the great challenges we have in healthcare is that radiology/cardiology/GI/pulmonary/Ob-Gyn images are not easily sharable between organizations. Although DICOM is a generally accepted standard, there is not an easy to use health information exchange in most communities to send DICOM data from place to place.

Sure, we could engineer numerous point to point solutions i.e. one organization's imaging modalities push DICOM images to another organizations image archive. However, such an approach is complex. Who owns the medical record? How long should the image be retained? What security should be used to transmit the image between organizations? How should the patient be uniquely identified if we want to storage all images for a given patient from multiple institutions together?

Here's the solution that all the companies entering this marketplace are suggesting:

1. A cloud computing offering is made available on the web for image exchange. No IT department has to host data, manage images, or deal with security issues.

2. A DICOM image is sent to the cloud along with a list of email addresses of individuals who can access the image and the length of time they can access it i.e. 30 days

3. The named individuals receive an email along with a URL and password to access their images on the cloud. For example, such individuals could be a consulting radiologist or cardiologist who may be over-reading the study.

4. A universal web-based DICOM viewer enables the authorized individuals to view the image without installing additional software i.e. there is no need to run proprietary workstation software from GE, Siemens, Philips, Agfa, Kodak etc.

5. After the reading is done, the username/password used to access the cloud computing site expires after the duration specified.

This solution is appealing for image sharing because it is low impact on IT departments, requires no local storage, is secure, requires no special software, and is simple to implement.

I look forward to many product offerings for healthcare imaging sharing via the cloud over the next year. That's cool.


Joe Lavelle said...

Great post John! I agree totally with you on the promise of the "exchange in the cloud".

Further, I believe the same concept should/will be applied to all results in an EMR not just DICOM images.

In fact, I do not think the Obama administration's plan for EMR will be successful until it includes:
- single national EMR system
- exchange of EMR info in the cloud

Automating the flawed process of having our paper medical records reside in each provider's (physician practice, hospital, lab, etc) office will be boon to EMR vendors but a waste of time and money for we the tax payer.

Moving our medical record to the cloud will be totally disruptive to health care entities at all levels (especially competitive health systems that believe they "own" patient data). But it is the only viable/maintainable long term solution.

Joe Lavelle
Managing Partner, Health Delivery Consulting
Santa Rosa Consulting

GreenLeaves said...

I agree that using cloud computing could address a large portion of the image “transfer” need. But from experience, the devil is in the workflow. Radiologists depend on comparisons and much cost is spent on balancing the display characteristics of the monitors so that the past image can be display next to the current image. To achieve usability, the cloud image will need to display along with the local image on the same workstation using a common set of tools. Another issue with the universal Dicom viewer is that up till now free viewers have had limited tool sets. To be acceptable, this browser will need capabilities similar to the open source Osirix available to Apple users.

The objective is to help the patient and many times it is beneficial to incorporate the image in the patient’s record. For these instances, the image needs to be transferred. The transfer security can be done relatively simply with secure FTP or public key encryption. On the receiving end, policies and procedures must be in place to identify the image as being from an outside source, incorporate it in the patient record and ensure that it is not copied or forwarded without express authorization. In the case of emergency rooms transferring studies to a trauma center, VPNs can be used to push images between Dicom nodes but this requires all the maintenance of point-to-point communications.

Long term I do believe we need to set up a protocol for initiation, validation, transfer and incorporation of Dicom data using a secure model of or

Marc said...


John Seely Brown and many other people use the term service grid which I like (although Cloud is pretty sexy). How do we balance easy access, which will promote more "meaningful use" with privacy and security. I would really like to leverage web services that exist today today but as you've mentioned I'd be breaking a whole bunch of laws if I did. Larry and Sergey surely won't sign a BA. Seems like the people protecting us are also blocking our sprint to the goal line.


The EHR Guy said...

I have been discussing with IHE members for defining profiles for the use cases that may arise from this approach or technique that you mention. I will be submitting a brief proposal soon.

Currently, I am rendering consulting services to a company on Computer Aided Detection (CAD) and other advanced image processing algorithms interoperability.

One of the ideas we have been pondering on is a web based service that can receive DICOM objects, process the images, and then return the results.

One of the main obstacles is the size of the images, especially when we talk about CT or MRIs and not to mention the pathology ones.

Still, getting the images to be viewed by the radiologists is a huge challenge. Most follow very strict workflows and adding another image to it is considered disruptive. Also, to integrate anything new with PACS viewing workstations takes a considerable amount of custom development and time, which can span from a month to over a year.

The current PACS viewing workstations would have to evolve to accomodate this type of technology that is being introduced. We know that with FDA firewalls introducing new technologies in this environment is time consuming and disheartening.

The musical chair approach to radiologist workflows is not well accepted.

Only a specific sub-domain, breast-imaging, has been able to use integrated viewing workstations but these are generally limited to a small number of radiologists in a given organization.

Introducing a new viewer to the radiologist is extremely challenging. For years I have been integrating different viewers and haven't been able to please all radiologists. Just a slight difference in Window/Leveling can discourage the use of another viewer.

For referring physicians this is not a problem since they use the viewing tool as a secondary step.

My two cents.

The HL7 Guy

Deborah Kohn said...

I agree with Dr. H about using cloud computing. But I also agree with GreenLeaves about the workflow issue and universal, clientless DICOM viewer.

For a few years, I have been working with one PACS (e.g., the archive) for all the "ologies" -- knowing too well how difficult it is to manage the completely different "ology" workflows.

Unknown said...

John: I see remnants of grass in your yard. Please consider the following low maintenance. Cheers.

Maimónides said...

Dr. Halamka, do you heard anything about Medting ( Perhaps would be a good solution.

PACSFerret said...

Collaboration in the cloud - check out Google Wave - if anything has potential that does.

Unknown said...

Great work man,i like your blog,its really interesting and enjoyable.

best wrinkle creams

Andrew said...

Interesting post, and an interesting direction for DI solutions/sharing.

However I think the questions you raise in terms of a complex approach are key and the highlighted suggestions from the cloud companies do not really address these. Instead they seem to address traditional infrastructure issues such as hosting, data management, security, etc.

Security can be assured (in email?) but someone at the consumer/provider end will need to own & monitor the cloud service for SLA compliance resulting in the same responsibilities of data management, identity management, auditing, privacy, etc. raising their heads again.

That being said, moving the yardsticks forward on infrastructure solutions in terms of availability and capacity particularly in this data-demanding field is a good thing.


Brent said...

Great post. Intelemage offers a service that fits well with what you are describing (

Unknown said...

Great post. Dicom Grid offers a cloud based service for medical image content management, where imaging studies are transported on the public internet and archived in cloud storage. Clearly this is an important aspect of digital healthcare.

Unknown said...

Great Post.

Dicom Grid ( offers a cloud based service that allows medical imaging studies to be shared over the public internet and archived in cloud storage.

Clearly the healthcare industry needs this class of solution to address the complexity and cost of legacy solutions.