A number of years ago, I was at a national conference in Israel regarding the management of digital x-rays, CT’s, MRIs and any other types of medical imaging. The technology that handles the collection, long-term storage, and retrieval of such images is called PACS [picture archiving  and communication service].

The principal behind this technology is actually very simple. The idea is that  various modalities [which is a collective term that refers to the devices that generate the images, like the x-ray machine and CT and MRI] transfer their information to a central computer that runs PACS software. The PACs software can then do any number of things. It can send the images to a long-term archive for backup purposes, it can reroute the images to another location for viewing by a senior radiologists or other specialists, it can process the images in a certain way so as to make them even more informative [like the construction of a 3-D model of the persons anatomy based on the collected information from the individual CT “slices”] and more. Once the information and images are all stored with in the PACS system, a physician working on the floor can retrieve these images in order to make a diagnosis and plan for further treatment for the given patient.

In the technical world, one actually refers to this process in the less attractive way of “garbage in, garbage out”. This reference is not meant to minimize the significance of the content that is being managed. Rather, the intent of this phrase is to focus the IT  developers on the principle that the  key function of a PACS system is ultimately to perform basic functions: collect, store and then retrieve the stored information for use by the physicians.

When presented in this fashion, it sounds as if PACS software should be very simple to design and as such should be very inexpensive. In practice, many professional PACS systems cost many many millions of dollars to implement and maintain. Why is this so?

The simplest answer is actually quite intuitive.. Imagine a juggler, juggling three balls. As difficult a task as this may be, there are many people who can do this effectively. Give the same juggler 10 items to juggle, where each item is of a different size, shape and weight, then the chances of failure are tremendously increased. In this day and age, PACS systems have to deal with tremendous amounts of images flowing into its data stores in shorter and shorter periods of time. In a large hospital, a single PACS system might need to intake multiple CT’s and MRIs and angiographies and plain films, all effectively at the same time. Rejecting any one of these studies, means that the original sending device [like one of the MRIs] is effectively off-line until it has managed to fully offload all of its data. More so, if there is a communication problem for any reason that interferes with the quality of even one of the thousands of images transferred, the entire study may need to be resent. This takes even more time and once again can delay the further use of the MRI for the next patient.There are solutions for such issues like “caching routers” and the like, but I will skip this for now.

The obvious question is why can’t an MRI or CT simply locally store all of the images until the PACS is ready to receive them? This is an excellent idea and one that has been raised many times. The problem is that it would require a fundamental change in the design of the MRI and CT and other modalities. Suddenly, these devices which are focused on capturing images, would now have to become mini PACS systems themselves, storing images until the main PACS system is available. What happens if there is a glitch or a power outage and the local copy of the images, that were temporarily stored on the MRI, is lost? The answer is that there is no answer and the patient may very well need to have the study repeated. One can already imagine how difficult this may be if the patient has difficulty mobilizing and/or is connected to various support devices and so on.

Without belaboring the point, although PACS functionality appears to be relatively simple and straightforward, in real life, there are many technical hurdles that must be accommodated in order  to have a smooth, secure and reliable PACS system working 24/7.

I will address one further issue which again seems horribly trivial, yet is a continuing problem in many medical centers. At some point before the imaging study is done, the modality needs to be informed as to who the patient is. This can be done in a number of ways. One way is to simply type all of the information into a blank form that appears on the screen of the CT or MRI or any other modality. There is a basic rule in data collection that any time information is reentered by a human, there is a significant chance for error. Therefore, retyping all of this information stands a good chance of leading to an error in the patient’s name or identity number or date of birth and so on.

Another option, which is very commonly used, is to have a list of all patients with appointments for the, say, CT, appear in a window on the screen of the device. The technician then only needs to scroll through the list and pick the correct name. This works well unless on that particular day, there are four patients with CT appointments, who all have the name John Smith. In such a case, the technician needs to use some other form of identification [such as the patient’s age or gender or identity number] to select the proper patient. There is also the issue of “fat finger syndrome”. This simply means that the user’s finger is wide enough such that it easily covers two or more of these selections on the list. Depending on which of these selections received the most pressure, the wrong patient can easily be selected.

Another option is to have the patient where some form of digital identification tag that is recognized by all of the various modalities. One such technology is called radiofrequency identification or RFID. An RFID can be programmed with the specific information of the patient. After that point, the technician need only wave the RFID over the RFID reader on the modality, and the correct patient is selected. Of course, using RFID, means that the entire hospital has to be reequiped to handle this means of identification. Every modality would need to have an RFID reader. Operating theaters that take advantage of this technology would need to have multiple RFID’ readers installed throughout every room. With the appropriate support software, an RFID system could make sure that the correct patients was being operated on and that the correct operation was being done. While the solution seems almost magical in its capabilities, it still requires a significant investment in time and money to fully implement. I could say that money should not be a factor when one has the opportunity to so dramatically improve quality of care. But unfortunately, I have found myself too often a lone voice when arguing that profits should be allowed to drop, if the benefit is better healthcare. And I will not further discuss this issue.

it is astonishing that technologies already exist that could so  dramatically improve quality of healthcare across the country, if not the entire world. But, if there is insufficient profit in implementing such technologies, they stand a poor chance of ever seeing the light of day. As a physician, it pains me to think of the number of people who needlessly suffer for the lack of commitment to one of the primary tenets of medicine, “first do no harm”. I have no magic answers. I cannot explain how people spend billions of dollars on pizza but cannot find the funds to support a system that could spare them needless pain, suffering and permanent disability. I hope that there is someone out there who is far smarter than I am, who can find a practical solution for implementing such technologies.

Thanks for listening