Wednesday, July 1, 2009

Sampling Charlotte at SIIM

The Taste of Charlotte festival held during the 2009 Society of Imaging Informatics (SIIM) meeting was an apt metaphor for this year's conference. The festival, held in downtown Charlotte, NC, allowed people to stroll by and sample a variety of foods. The SIIM meeting allowed attendees to taste a little bit of new technologies and configurations from diagnostic imaging informatics vendors. It was also a chance to network with colleagues, old and new, as well as make new contacts. 

On a more somber note, the thunderstorms outside on the streets of Charlotte were reflected in the mood of vendors concerned about attendance (down from last year), the overall economic situation, and the potential impact of new healthcare regulations. 

For me, the highlight of the conference was the PARCA meeting, where a good dialogue took place between PARCA and the American Board of Imaging Informatics (ABII). There was a healthy discussion by members that examined the pros and cons of the two prevalent certification authorities for PACS administrators. 

The advantage of PARCA’s on-line exams was a clear favorite, as was the organization’s global reach that is reflected in its membership. Both organizations are currently revaluating recertification options. I hope the respective board members will have the vision and willingness to continue their dialogue to facilitate a single certification organization for PACS administrators. 

I also found the updates on digital mammography and pathology to be illuminating. 

In regards to digital mammography, it was clear that most PACS vendors are still far from being able to provide a decent functionality for full-field digital mammography (FFDM) on their workstation. In turn, the modality’s mini-PACS and its dedicated mammo workstations interacting with the main PACS creates several interoperability issues. 

Regardless of informatics challenges, digital breast imaging has a promising future, with tomosynthesis, breast MRI, nuclear breast imaging, and thermal breast imaging already here. However, those informatics challenges are by no means minor issues. Sorting through image sets, developing hanging protocols, and other challenges mean that it will be a few years before the full complement of digital breast imaging will be part of the mainstream. 

Digital pathology was scary to the degree that the required data storage is humongous. The typical weekly data generation is equal to what a radiology department might create in a year. Similar to the challenge of a radiology workstation having to beat the performance of a film alternator, the pathology workstation has to simulate and beat the performance of a pathologist using a microscope, which will be hard. The early implementations will be very likely telepathology in areas that do not have access to laboratory expertise. 

Next year SIIM will be in Minneapolis, which is a beautiful and vibrant community. However, I hope the site-selection committee will select more visitor friendly locations such as San Antonio, Vancouver, Seattle, Boston, or even Dallas in the future. It is always nice to have options to network with colleagues at places other than a conference hotel lobby. 

DICOM Structured Reporting, Part 1 of 2

Part one of a two-part series on DICOM structured reporting.
CT scanners are being equipped
to export dose information
 in a Structured Report

There are basically three different types of options when it comes to diagnostic imaging reports. The first option is traditional reporting—the dictation of interpretation by a radiologist that is then converted to text by a transcriptionist. The radiologist will then read, proof, and edit the report before sending it to the ordering physician. 

The second option is the use of voice recognition (or speech recognition) technology. A radiologist dictates their interpretation, which is converted to text by the application. The radiologist then proofs, edits, and sends the report to the referring physician. 

Voice recognition technology commonly employs macros, or canned report formats that a radiologist will utilize to speed their workflow. For example, rather than recreate a normal chest x-ray report, a radiologist will create a macro for this interpretation and select it when appropriate. Many radiologists say that voice recognition with macros helps them increase productivity and allows them to be much more efficient. 

Structured reports are the third option. The structured report can be a macro, but it also contains structured contents, structured measurements, codes, and so on in a predefined format. 

A DICOM structured report, among other features, allows measurements—such as those taken during an ultrasound examination—to be captured at the modality and then exchanged with a PACS. It creates a DICOM object with the same format as a DICOM image. It has a header, patient information, patient name, ID, birth date, sex, and so on. But instead of having an image, it has the information about the measurements. 

What is the major advantage of using DICOM structured report for measurements, such as from ultrasound? Part of the radiologist's workflow involves cutting and pasting, or copying, this information into their reports. 

The PACS can interpret the DICOM structured report and automatically fill in measurements in the voice recognition or the transcription or the reporting application. In addition, the PACS can take the DICOM structured report and export it to an electronic medical record. 

Objects in DICOM are expressed as IODs, information object definitions and they reside at the series level. So, when viewing a DICOM study that contains a DICOM structured report, you would see two series: Series 1 would be the images; and Series 2 would be the structured reports. It would be identified by the tag "SR" for structured report. 

DICOM structured report objects behave like regular DICOM objects such as images or presentation states. They can be sent, queried, retrieved, and are handled like any DICOM object. They have a pointer that references back to the original object where the measurement or the observation or the information about the image is contained. 

The contents of the structured reports are highly encoded, which means there is a code for everything. For example, there are codes for measurement units, whether it is in centimeters, millimeters, or inches. There are codes for the observers, the diagnosis, and the procedures. These codes provide an excellent mechanism for extracting information such as for outcome measurements or data mining. 

What are some common applications of DICOM structured reports? In addition to measurements, a physician can use it to identify key images as well as CAD observations. Outside of radiology, the cath lab has a lot of opportunities for structured reports such as the image groups, ECGs, references, reports, and so on. 

A relatively new DICOM structured report includes the mechanism to exchange dose information. Healthcare providers are concerned about the quantity of radiation dose that is administered to patients, particularly with the advent of the multi-slice CT scanners. New regulations require the registration of the dose of any examination that exposes a patient to x-ray radiation, starting with the CT exams.

How do we get that information? 

Well it’s not sufficient to have the information taken from the image header, because when you have a flouro study, one might do quite a lot of observations, but not create any images. So the record in the images would not reflect the actual dose that would be administered to the patient. 

The MPPS, the modality performed procedure step-- a mechanism to allow exchange of information between a modality and RIS and PACS--has a place for dose information. However, this is not necessarily sufficient because if images are not generated, there might not be an MPPS. 

Using a DICOM structured report for radiation dose is highly preferable and most, if not all, new CT scanners do support this option. This allows for the information to be exchanged among all ionizing radiation producing modalities as well as the PACS. 

An important element to note is that there is no presentation information in DICOM structured reports. So when you get a DICOM structured report, there’s no rule about what should be on top, what should be left, what should be the font size, what should be the header, what should be the footer. This is an opportunity for vendors to deliver a consistent and readable report for the user and allow them to configure it to their preferences. 

The DICOM structured report has a much higher data integrity, which simply means that information which is captured once and digitally communicated alleviates the potential for error from copying or re-typing that data. It is also much more efficient and has the potential for delivering a higher quality of patient care for its adopters.