logo PhoneClick to Dial 720.560.9785 MailClick to email inClick to Linkin
back to homepage

What to look for in selecting a PACS performance monitoring system

 In the last 5 years,  PACS has without doubt, has reshaped the world of diagnostic imaging and Radiology.  Before the PACS systems were being used, Radiology and all downstream health care providers dependant upon radiological diagnostic interpretations, have depended upon the use of celluloid film for images to be read by the Radiologist. Before images in electronic form were used,  films were exposed, developed, hung for display and stored in vaults. The time to get the films to the radiologist was on the order of minutes to hours.  With the advent of PACS, there is a new expectation by the radiologists and other users of PACS to have images within a few seconds after a computer’s mouse click.     

This expectation of fast image delivery was an over night shift in work flow for everyone in the Medical Imaging support chain. With these new expectations of fast image delivery firmly in place and driving the Radiology department work flow, support personal of the PACS system and others responsible for delivering medical images to radiologists and other providers in the patient care chain, have been scrambling for solutions that will empower them to help maintain fast image delivery.  In order meet the new expectations the support staff needed a method to maintain, verify and alert support staff of the performance this new electronic media.    A pro-active, real-time system that continually monitors the PACS system would be ideal.  If it alerted on the most relevant issue in Radiology department efficiency or plainly - when slow-downs occur in the electronic image delivery chain, that would be the best. 

  In the recent past, there have been several new companies pioneering solutions that are aimed at helping the support staff monitor PACS performance. For those of us not involved in the day to day details of a PACS system, it seems reasonable to monitor everything one can to maintain a PACS system and to that philosophy some have created software solutions that monitor transactions involving the PACS database, bulk flows of network traffic between server, disk drive capacities and other tertiary and benign elements in a PACS system related to it’s core performance.  The reality of how a PACS functions, is that unless the PACS database is completely down, rarely do transactions take more than a few microseconds to milliseconds to complete.  As any PACS system administrator will attest to, the Radiologist using the system doesn’t perceive small database related delays and unless the database has crashed, they never will perceive a database delay or any element not directly affecting the timely delivery of images. Bulk network traffic monitoring tells nothing about when the Image traffic is traversing across the network.    Unless storage is running very low, or no available storage at all, due to very poor storage planning disk drive gas gauges also tell nothing about how the system is performing.
Other elements some software vendors are offering as mistakenly related to a PACS system performance are how many users are currently on a database or how many anomalies are in a queue and how many transactions are open, or even if they display when a modality is sending in a series of images to PACS. All of these elements are not capable of delaying images and have really absolutely nothing to do with image delivery performance. It can’t be mentioned enough, that mage delivery performance is the most perceived detrimental issue in a PACS system. It also affects the work flow of the users the most. 
To understand what are the real factors in image delivery efficiency, we need have a clear understanding of the elements affecting the delays of image delivery.
In a PACS system.  We have ‘edges’. Simply stated, and edge on one side is where the images are created and the edge of the other side is where they are viewed. On the creation edge, the images traverse from the CT, MR and Ultra Sound and all other DICOM based modalities to the imaging server where they are forwarded on to the SAN (Storage Area Networks) or Storage Grids. Through user based queries, the storage system then forwards the images from storage though IIS or image servers, content switches, via network switches and routers to the user workstation on demand. On the created edge, images can also be delayed from the computers sending images to storage at the modality through poorly configured Network card settings caused by the vendor upgrading their software. A phenomenon that happens almost all the time and with every vendor.
At any point in the middle of a PACS system, images are electronically riding on a network. Images can be delayed for many reasons and most delays are unwelcome and happen both unannounced and until now, usually undetected.  Images are large quantities of large files and always a continuous stream of traffic. Moving images around a network from modality to Storage and on to the workstations is the largest workload for the hospital network. As any network engineer will attest, image traffic easily dominates a networks usable bandwidth.    
Unless on its own dedicated network system, Image traffic is also easily susceptible to delays caused by competing network traffic.  Other factors causing image delays include -incorrect network equipment configurations, defective firmware and software updates and a host of other things that can and do go wrong all the time.
On the other edge of a PACS system,  hidden delays can also be introduced at the user’s workstation by incorrect anti-virus or other software settings the user may have the rights to adjust. Users loading toolbars and other applications can also effect the workstations performance.   As the PACS system delivery is really a single pathway in to storage servers and out to the users, any element in that pathway slowing down will impart a slow down in the delivery of images to the users. It is just simple physics.
A PACS systems behavior will always be controlled and best understood by the weakest link in the chain theory and thus it is vital and most effective to monitor it from one edge to the other and at each link along the way. These are the most important parameters to look for in any PACS system performance software.
To do the best job at performance measurement, we must have software capable of monitoring the factor that not only is seen by the users the most but what also effects the workflow of the Medical Imaging department the most.    
So in our quest to determine what is the best PACS performance monitoring software system, we now have a good idea that it needs to measure, monitor, alert and report on when the images are traversing slowly.   What is slow?.  Good question.  Everyone’s perception of ‘slow’ is going to be different.  Human perception is effected by many things but as all things in nature, human perception falls under a standard bell curve varying from very sensitive individuals to those that image flow needs to come close to complete failure or completely stop before it is noticed.
Another desirable trait in a monitoring system is the ability to be alerted  before the users register a slow-down complaint. Through flexible intelligent threshold adjustments, it is possible to adjust when an alerting email arrives before the phone call from the users does and most importantly and before it effects patient care. Thus the performance monitoring and alerting software needs to be flexible and allow the users to learn and adjust where the alerting threshold is set. Many do this but as mentioned earlier, it is up to you to choose a vendor that is monitoring and alerting on the right parameters. Not parameters having little to do with true system performance.

  So the most desirable traits in a PACS performance monitoring system is now a team of features which is centered around the element most effecting the users workflow and causing downstream delays to patient care.    It should be clear at this point that the most important thing to monitor and most effected by slow-downs is the image traffic as it traverses through the PACS system.  

  Authored by,

Michael Lunsford
PACS IT -  Imaging Systems Administrator
Exempla Healthcare
Wheat Ridge, Colorado, USA