Back to Healthcare Contents
Back to Healthcare Contents
Electronic archiving of medical images from a busy department over many
years will require dozens of terabytes (TBs) of digital storage and rapid
access to all the images. Several technologies are available today to store
large amounts of image data. However, there are many tradeoffs and architectural
considerations when designing an archive system.
Recent advances in storage technologies have drastically reduced the cost
of storage as well as improved access performance, reliability and capacity.
There is no longer a need to wait for a cost effective archive solution with
the capacity to keep your entire medical image archive accessible. Automated,
Nearline archive systems today cost fractions of a cent per megabyte and
provide virtually unlimited capacities. Systems range in capacity from one
TB to hundreds of TBs, enough for even the largest departments and hospitals
to keep decades of studies and patient records on-line. Nearline is a StorageTek
trademark and was invented by StorageTek in 1987 with the shipment of the
first automated robotic tape devices that bridged the gap from on-line storage
and off-line (on the shelf) storage.
Radiology professionals understand that a long-term archive is the
foundation of the filmless plan; it is a key to the successful implementation
of a radiology or cardiology PACS system. The long-term archive must provide
immediate access to your entire image archive without human intervention.
The challenge, then, is to implement a cost-effective, reliable, high-capacity
archive solution. Historically, facilities have left the critical archive
component of the system until last because an effective solution was unavailable.
Today, leading modality vendors, digital film system providers and PACS
integrators rely on StorageTek to deliver the archive that, when coupled
with their software, meets the following requirements:
There have been very dramatic decreases in magnetic RAID disk and automated
tape costs over the last few years due to significant spending in research,
development and manufacturing in these technologies. Optical is decreasing
as well, but not nearly as fast as tape or disk. Pre-fetch and ad-hoc performance
of magnetic tape systems have also improved dramatically in the last two
years.
Medical images usually have a very definite retrieval life cycle.
New images require immediate access by several people in many different places
within a medical facility. Access to a critically injured person's x-ray
within minutes, if not seconds, can be a key factor in the diagnosis and
treatment of the patient. However, after the patient has been treated and
discharged, the same x-ray may rarely be accessed again. If it is, minutes
or even hours to retrieve this x-ray are normally acceptable.
The requirement for fast retrieval of current studies, then slower retrieval
for historical studies, suggests that different types of storage devices
could be used over time to store an image. As fast retrieval times grow less
and less important, the image could be migrated to less costly storage devices
to minimize the cost of storage over the life of the image. Figure 2 illustrates
this concept.
Total cost of storage in this example (in units of dollars per MB per
year) consists of purchase price and ongoing costs such as maintenance, removable
media costs, power/cooling and floor space costs. If an organization considers
all of these costs of over a 7 year period, an annual or monthly cost per
MB can be easily calculated. This amount is the vertical axis in Figure 2.
The area under the step curve is the total cost per MB to store data over
time. The area under the step curve can be minimized by decreasing the cost
of each storage step and by shortening the time an image is stored on more
expensive storage levels.
As images occupy space on RAID disk, it costs more per day to store than
it would on optical, for example (notice that the step curve is not drawn
to scale in this chart). Therefore, the sooner a study can be migrated to
less expensive storage, the less it costs to store that study over time.
The automatic and efficient migration and management of the studies in the
archive is easily done today with a software product like STK's Application
Storage Manager.
One of the major requirements for medical image systems is to provide
the radiologist with immediate access to as much information as possible
on the patient. The concept of pre-fetching archived images before a patient
arrives in the department is one way of providing historical studies to the
radiologist quickly. This requires an interface to a patient scheduler. With
pre-fetching, you can now "mask" the relatively slow retrieval time of the
long term archive from the radiologist and still save significant amounts
on long term storage.
Figure 3 is the output from a performance model comparing four removable
media archive technologies: 5 1/4" MOD optical, Digital Linear Tape, and
STK's Timberline and Redwood automated tape products. The size of a file
stored on removable media has a dramatic effect on capture and retrieval
performance. The figure plots file or study size (in MBs) vs. the number
of studies per hour that can be retrieved in a pre-fetch run.
For file sizes up to about 50 MBs, all four technologies perform fairly
evenly, but as files get larger or multiple files are retrieved per mounted
piece of media, magnetic tape quickly becomes the highest throughput device.
This performance gap is because of tape's much faster data transfer rate
of up to 11 MB per second compared to optical's 1 to 3 MB per second rate.
For example, at 600 MBs retrieved from the media (15 radiological studies
or 2 cath lab studies), Redwood and Timberline can transfer six times more
information per hour than optical. Digital Linear Tape transfers about three
times as much as optical. This performance gap continues to widen between
Redwood and optical due to Redwoods 11 MB per second transfer speed.
As new digital modalities become accepted and more pervasive in the coming
years (CR, DR, MEG, Functional MRI, and mammography), radiological studies
will grow from an average of 25 to 50 MBs to a few hundred MBs. Cardiology
studies today already have file sizes of 250 MBs to 1,000 MBs with Cath Lab
and echocardiograms, so tape fits extremely well in this environment due
to its high data transfer rates.
Compression will obviously affect the archive capacity and therefore cost.
All the information here assumes no compression. If lossless or lossy compression
is considered and implemented, each archive technology scales equally smaller
and there is little or no difference in the cost between these technologies.
With the low cost of Nearline magnetic tape, compression becomes a "non-issue"
from the archive perspective. Uncompressed, the archive cost is well below
a penny per MB, with compression it is even lower.
In almost every article published on optical storage devices, you
can find claims of media life of 25 years, 50 years, even 100 years. This
long shelf life is nice, but in today's rapidly changing high tech environment
the question must be asked, "What is involved in actually retrieving and
reading that piece of media in 25 years?" Even with today's CD "standard,"
how long will you have a CD drive and the appropriate software to read the
information off the disk? With DVD promising much higher capacities and higher
performance, CDs may quickly go the way of 5 1/4" diskettes. Do you have
a PC today that has a 5 1/4" diskette reader in it? How about an 8" diskette
reader?
It appears now the DVD may consist of several standards. How long will
the DVD drives support reading of "old" CDs? The original DVD specification
did not even provide for reading of existing CDs. Fortunately, the vendors
developing this technology realized the importance of providing backwards
read capability for at least one previous generation of CDs.
Automated archival tape systems are now a very robust, reliable and very
high capacity storage medium. Media life can be 15 to 30 years with metal
particle tapes that are common today, with tens of thousands of loads and
read/write operations occurring with no problem. If a "soft" error ever occurs
on the media, error correction routines detect this and automatically correct
the error without host computer or operator intervention, ensuring that the
data will always be readable.
The main point here is to consider and plan for technology obsolescence
over a reasonable period of time. This will mean that sometime in the future,
whether it is 5 years or 15 years, you will want or need to take advantage
of some quantum leap in storage technology, and this may mean a media conversion
effort. This media conversion effort has been done before with relatively
little pain and cost if planned for and done correctly.
In conclusion, an effective, low cost, long term archive is critical
to the successful implementation of a PACS system and any filmless plan.
Performance and capacity requirements must be carefully understood and
cost/performance tradeoffs must be made in light of these requirements. Many
hospitals have recently implemented high capacity archives and are now reaping
the benefits of access to all of their images, without human intervention.
For multi-terabyte medical image archives, automated Nearline archive systems
are the only economical way of storing massive amounts of data required for
this application. Storage management software is also a critical component
of any medical image archive, enabling automatic management of the archive
and assurance that the cost of the archive will be the lowest possible over
time.
Since 1969, StorageTek has stored, retrieved and transferred
some of the world's most critical data. Our products have proven to be highly
reliable in the most demanding environments.
StorageTek reported revenue of more than $2 billion in 1996. Our customers
fill the ranks of the Fortune 1000, government and the public sector, and
comparable international enterprises. With headquarters in Colorado, StorageTek
sells, manufactures and supports equipment and software worldwide.
StorageTek's web site is located at www.storagetek.com.
Tim has been in the "high-tech" industry for 20 years. He
was employed at IBM for 12 years where he was a development engineer, systems
engineer then a marketing representative. He has been with StorageTek for
5 years now in various marketing and program management roles. He has degrees
in Mechanical Engineering from the Univ. of Southern California and a Masters
in Business Administration from the Univ. of Colorado. He currently is the
Medical Systems Application Manager at Storage Technology Corporation. You
can email him at tim_chunn@stkamer.stortek.com. StorageTek's web site is
located at www.storagetek.com.
STORAGE AND RETRIEVAL PERFORMANCE AND COST TRENDS
Figure 1 illustrates the dramatic decrease in cost of three types of computer
storage technologies over the last several years as compared to paper and
microfilm medias. Radiological film is even more expensive due to the larger
size and greater storage costs compared to microfilm. Most people do not
realize that due to increasing costs of labor and floor space, traditional
manual methods of storing and retrieving information now cost as much, if
not more, than much faster computerized methods.
THE MEDICAL IMAGE LIFECYCLE
CAPTURE AND RETRIEVAL PERFORMANCE
DIGITAL ARCHIVE LONGEVITY
THE "A" IN PACS IS HERE