Monday, February 14, 2011

Medical Imagery Bandwidth Solution

A series of questions need to be answered before you can really get close to a "final" solution addressing medical imagery bandwidth.

First question you need is to determine your scale. Some typical questions are:

1) Average File Size to be sent
2) Number of files sent / time (day, hour, min)
3) Time Critical Nature of transmission (Immediate, Few Hours Late, Day Late)

This will give you an understanding of your bandwidth needs both from a Max Size and Total Volume standpoint.

After you've determined that, you need to outline the geographic nature of your sites. Are they regional in nature or national. This will help identify whether a Local provider (ie Cable/MSO or Small CLEC) or a National provider is a better fit. Local providers are usually less expensive, but you run into challenges in using them nationally (b/c they are usually going to have to use 3rd parties for the backhaul/transport) especially when it comes to outage management. Local providers are often less likely to have systems to verify SLA'a and be able to give you an end-to-end view.

Next set of questions to be answered relate to the capabilities of the IT organization supporting these sites and the CPE. If they are bare bones, then you want either a relatively simple solution (ie Ethernet focused). If they are more advanced, then a TDM solution may be acceptable.

Finally, once you've determined the answers to the above, now you can consolidate those into a set of requirements for bid. My recommendations follow for 2 scenarios:

Case 1 .....

- Medium Sized Bandwidth needs (<10MB ave file size, 100-200 files/day, hourly delays acceptable)
- Regional Focused with limited IT resources.

Recommendation: Business Class Ethernet/Metro Ethernet offering from local Cable/MSO or CLEC

Case 2 .....

- Large Bandwidth Needs (500 MB file size, 1000's of files/day, real time transfer). - National Focused with well skilled IT resources.

Recommendation: DS-3 or similar Fiber Based Bandwidth solution from National CLEC or ILEC.

What I have found is that the best solution for any company has to be specific to their needs. How big the facility is, in the case of a hospital, how much of their records are currently and in the future will be digital, must they send MRI and X-Ray images out to other sites, etc. Do they use VOIP for their phone system. All of these can determine what the best solution for their specific needs. If they use a lot of bandwidth, like an average of 80 to 90mbps right now, then fiber optics is the way to go. If they currently have a single T-1 but project that in the next 1 to 3 years they will need to have 5 T-1's then I would recommend upgrading to a DS3. If fiber is available and easy to get access to, then that is always the best route, as it is far easier to scale it up to higher bandwidth needs than you would find with traditional copper (T-1/DS3).

Transmitting images over a network backbone is all about speed. If fiber optic is available in the area, that would be my first choice for medical imaging applications. Fiber is very scalable and the prices are can be dirt cheap compared to copper lines, such as T-1.

Although slower, the T1 lines come with an SLA (service level agreement) and regular line monitoring. You may need to check if the FTTH comes with a similar SLA.

High resolution image files are the the most common larges files used in medical imagery. Are these files being moved across a WAN? This is the usuall set up. What is the baseline usage for this link or links? What is the criticality for real-time viewing, in other words do you need this to be viewed by other persons as the images are being produced .... or can you move the images during less bandwidth intensive periods?

You can have as much bandwidth as what you want to pay for. But if you only need it to be at a new location the next day .... then your options are increased and become BOTH application and bandwidth based.

I would not bother beginning such a project without a realistic estimate for the amount of traffic the link(s) will need to support, and drafting SLA requirements. Data size, reliability/DR requirements, and SLA metrics will drive the technical choices.

In practice, if you need to link small clinics to a data bank ..... especially if the remote locations will accept best-effort link ..... than a "business" T1 should provide enough bandwidth for transfers.

If you need to link large sites or data banks, try as much as possible to get ethernet (metro) links .... if those aren't available at a minimum look for a DS3 bandwidth or Sonet solution (e.g. OC3 bandwidth).

Generally speaking .... the best method for medical imagery applications depends on volumes of information and timescales for transmission, storage, processing.

Any organization which spans multiple continents would do well to chose the optimum solution for each geography .... and either the operators interconnect via NNI or the medical organization can do the same (chose carrier neutral facilities to achieve this cost effectively).

The advent of VPLS raises the question as to who should offer / own the connectivity.

Because VPLS supports multiple Layer 2 logically separate interconnects, it is possible for medical imagery to be just one of many applications available.

Therefore other applications can run "for free" across the same connectivity. Patient management systems, GPs (local doctor surgeries) can easily be added ..... even interconnects with public telephony and video applications so that patients can be visited by friends and relatives from home.

It will be interesting to see whether the medical imagery networks that are extending outwards also get used for other applications. Or whether networks for patient information and similar applications get upgraded to also support the medical imagery.

The answer is going to depend largely on the type of image. Each image type has different characteristics. The worst 'offender' in terms of bandwidth will be cardiology type images which offer essentially full motion video of th heart at very high resolutions. Secondly, it depends on what is being done at the site - diagnostic reading requires the absolute highest resolution where as consulting can often be done through a web server with lower resolution.

Finally, understanding reading patterns are key. In a large organization where the reading may be done at a different location than the image acquisition, a cloud type of network may be desirable rather than a traditional hub/spoke.

The problem with PACS is that once users start using it, the usage will increase and the usage patterns will become almost completely unpredictable unless you have very detailed information on referral patterns within the organization.

In general, metro ethernet is a relatively cheap solution that can allow for scalable bandwidth. Some providers are also providing Metro E over DOCSIS which can be very appealing for small sites.

For help in walking through all of the questions, options, pros and cons .... saving you time, effort, and money ..... I suggest taking advantage of the free support offered through: Bandwidth Solutions

Labels: , , ,

0 Comments:

Post a Comment

Links to this post:

Create a Link

<< Home