Standards Watch | Multislice Challenges PACS

The new generation of higher volume multislice CT scanners is greatly challenging the management of images, the radiology infrastructure, and requires new standards to meet these challenges. That's no secret. Only two years ago, a four-slice CT system was state of the art. Today, facilities are adjusting to the volume of images generated by 16-slice scanners as well as 64-slices that are generated in a third to half a second timeframe. And looking forward, CT manufacturers have 256-slice CT scanners in their development labs.

A PACS is essential in managing multislice CT images. Without PACS to view and archive the large volume of images and data, one might as well forget the purchase of these devices. Among the considerations for effectively managing CT images are throughput, workflow and DICOM support. So, given that we have a PACS, the following challenges exist.

Throughput: Imagine that a complete CT study has to be sent from the ER to a radiologist - who can be in the reading room, at his home and/or office, or even in another city or another continent when nighthawk services are being employed. A high-speed network connection is essential. In addition, potential bottlenecks in the application software become obvious. If a typical study in the past was 200 images, which might have taken five minutes to transfer, a 1,000-image study will than take close to half an hour - which is unacceptable for trauma cases. High-speed network connections using compression are a requirement. Proper configuration of network hubs and routers is getting critical as well. At this time, you may find out that you do not run full duplex, or are sharing bandwidth with other users; it is time to work with your IT department and/or network people to make sure the communication "pipe" you think you have is indeed allocated properly.

Workflow: Several institutions have been using QA stations and gateways as part of their PACS to verify the demographic information and match the CT study with the radiology order. Imagine that all these images are sent to a QA station first where they are verified and then sent along. This might cause unnecessary delay, especially if we are talking about thousands of images. Thus, revisiting the workflow is necessary.

DICOM support: New applications such as CT angiography and others were not envisioned at the time the DICOM standard was written for CT scanners, more than 10 years ago. The "old" DICOM CT definition is based on the old paradigm, whereby images were treated as slices. These new applications are much more efficiently encoded as so-called "multi-frame" objects, whereby the slices are packaged in a single "volume." This has resulted in the definition by the DICOM standardization committee of a brand-new DICOM CT object, similar to what has been done for MR and is being done for cardiology and angiography. It should be expected that vendors will start providing these new objects very soon, and it is highly recommended for users to request the "Enhanced CT" DICOM support.

Image management: How to group and/or split the studies is something that many users wrestle with. The problem is that there could be a discrepancy in the ordering, acquisition, and reporting process of diagnostic imaging procedures, i.e. multiple orders can result in one or more exams, which need to be viewed and reported by one or more radiologists. This is particularly the case for a whole body scan which can cover multiple body parts, such as a head, neck, chest, abdomen and pelvis. These body scans are typically ordered with multiple procedure codes and corresponding Accession Numbers. In the past, when the number of images for an exam were relatively low, say 100 or so, it was manageable for a radiologist to view these on a workstation. However, a neuro specialist who is merely interested in seeing the head images might not want to have to download yet another few thousand images of the chest and pelvis. The handling of these "composite studies" can be handled as follows:

From an order perspective:

  • The radiology information system itself can generate a "master order," with a special, internal CPT code identifying a whole body scan. In addition, the results of this master order can be reconciled with the multiple requests within the RIS.
  • If an institution uses a broker, which translates the HL7 orders into a DICOM modality Worklist, the master order can be provided in the broker, i.e. outside the RIS. This allows a single entry in the Worklist, however, the reconciliation of the master order with the multiple requests later in the RIS has to be done separately.

From an acquisition perspective:

  • Modalities seem to prefer the IHE-defined PGP approach, whereby the different body parts are identified using different DICOM Presentation States. The IHE demonstration at InfoRAD at RSNA last fall showed some prototype implementations.

From a QA perspective:

  • Some of the QA station software allows a user to manually split the study or split a study into multiple Series by dragging or dropping the thumbnail images into a new Study or Series viewport. It should be noted that manual study split may not be acceptable in ER applications because it breaks the workflow and impacts timely availability of the images.
  • Some QA stations use splitting software for automatic distributing the images over two studies. Note that there has to be an overlap between the studies, i.e. several images would show up in more than one resulting split-study. There are a hand-full of installed systems that provide this functionality.

From a reporting perspective:

  • The challenge is to only make available to the physician the images he needs, not providing the potentially thousands of slices that cover the anatomy which is not being reviewed. The issue is not only making the appropriate images available, but also showing in the reading Worklist the different series and body parts so that the radiologist can choose appropriately.
  • It should be noted that there are institutions and radiologists who challenge that there is an issue, either because they do not sub-specialize or they want to see the complete study anyway. If they cover small institutions, a single radiologist will review the complete study anyway.
  • For ER applications, the radiologist also typically scans the complete study. In addition, some radiologists claim that they should have easy access to all the images regardless, even if they sub-specialize. However, I suspect that the latter comment is more a justification of not having a split-study feature.

From a pre-fetch perspective:

  • Eventually, the studies are very likely to be migrated to secondary archive media that may not have fast access. Pre-fetching or downloading thousands of not-needed images is an unnecessary burden on the infrastructure and could slow down the actual retrieval of the images that are of interest. Pre-fetching the appropriate subset can only be done by identifying the body part as a retrieval key.

From a review perspective:

  • Downloading the whole-body CT would be an unnecessary burden.
  • The solution to provide for a physician only certain images depends on the PACS architecture.
  • Some PACS have a separate web viewer that allows only for a certain amount of time for the images to be stored. One could push only a subset of the images to the web server.
  • Some PACS have an integrated PACS web server and archive, for which it does not make a difference.

From an order management (billing) perspective:

  • The radiology order has to be "completed" so that radiology can appropriately bill the patient and/or insurance company for the services provided. The challenge is to match up all the multiple requested procedures with the single performed procedure. The approach is often manual reconciliation and/or linking. Mostly, a system administrator closes out the requests that are satisfied by another acquisition.

For the above reasons, it is essential that a proper, well thought out plan be put in place to effectively manage multislice CT images. Facilities purchasing multislice CT scanners must prepare for the information requirements just like they do the physical structure of a new CT room. I have had calls from users wrestling with the workflow issues who are dissatisfied with image access, particularly to trauma cases, most often because they did not anticipate the necessary changes. If done properly, these devices are a true asset and will very positively impact patient care. As with anything else, proper planning and getting all parties involved from IT to radiology administration to the clinical specialists is key to a successful implementation.

Herman Oosterwijk is president of OTech Inc. (, which provides PACS seminars, books and computer-based training.