John C. Hayes

It's been a bit more than a decade since teleradiology really started to take off in the U.S. Fed by a radiologist shortage in the early 2000s and a desire by many practices to shed their night call work, teleradiology companies grew and prospered throughout the decade to the point where they are today: large and diverse organizations reaching into nearly all aspects of radiology practice, and also a potentially a disruptive force that could dramatically challenge traditional styles of practice.

Inside, beginning on page 14, you'll find a series of vignettes by some who have followed the development of teleradiology during that time. They are, for the most part, positive about the contributions teleradiology has provided to the profession. It's relieved night call pressure, but it's also provided expanded access to subspecialty interpretations, around the clock and geographically. For some radiology practices, teleradiology technology has been a way for them to expand their reach in their community and regionally.

As we've reported over the years, the decade brought some dramatic changes. NightHawk, then Virtual Radiologic became publicly traded companies. Venture capitalists roamed the RSNA exhibit floor looking for investment opportunities in teleradiology-and found some. Then Virtual Radiologic was purchased by a private global investment company and, in its turn, announced plans to buy out NightHawk.

The combination of NightHawk and Virtual Radiologic will create the largest radiology practice in the country, with 325 radiologists serving 2700 healthcare facilities in 50 states. According to Virtual Radiologic CEO Rob Kill, the combined organization will control a bit more than 1% of the $20 billion market for radiology professional services.

Most of the sources interviewed for our article were not worried about the acquisition, but those who participated in our online poll asking about its impact on radiology practice were not as confident. With 145 responses as of Nov. 9, just 18% said the impact would be positive. Another 55% said the impact would be negative. Eighteen percent said it would be neither positive nor negative and the rest were undecided.

Thus, while teleradiology has clearly become a part of the fabric of radiology practice, there is still some unease about where it's going to take us. One possible path is apparent in data compiled by the KLAS research group and released in October. Traditionally, teleradiology contracts have been negotiated through local radiology practices. Both NightHawk and Virtual Radiologic have affirmed their commitment to this approach.

KLAS researchers, however, found that hospitals and clinics hold close to 40% of the teleradiology contracts. One of the researchers was quoted as saying, “In several situations, teleradiology decisions are being made at the health system level.”

KLAS also observed that expectations are for greater teleradiology volume in the years ahead, and they note that providers are showing growing interest in receiving final reads from teleradiology companies. If this becomes pervasive, where does it leave the local radiology practice?

While Virtual and NightHawk have pledged to not compete with local practices for radiology contracts, we know other teleradiology companies are not similarly committed. Further, demands by those with an economic interest in radiology services have a way of changing even the most altruistic of attitudes.

Combine this with a desire by hospital executives to wield more control over imaging, and you could see a decade ahead in teleradiology that's even more eventful than the one just past.

Articles

It's been a bit more than a decade since teleradiology really started to take off in the U.S. Fed by a radiologist shortage in the early 2000s and a desire by many practices to shed their night call work.

After a decade of growth in teleradiology; now what's next?

Do you think stripping out textual identifying information in publicly available radiologic images will protect you against privacy violations? Think again. A paper presented Thursday at the 2010 RSNA meeting showed how facial images reconstructed from maxillofacial sinus and cerebral vasculature images could be matched in a database using commonly available face-matching software. 

The study mixed 29 faces developed using radiology postprocessing software with three sets of facial databases and then used a face-recognition program, Picasa 3.6, to look for matches. The software program hit the mark 27.5% of the time in each of the databases, said Dr. Joseph Mazura of Weill Cornell Medical College.

“For a match to be accurate, two things needed to occur. Picasa must recognize an image as a face and then must match it with the with the subject’s photograph,” Mazura said. In all phases (the databases were expanded 50 images at a time, to include a final total of 179 with the reconstructions), reconstructions were consistently matched or not matched at a 27.5% rate with an upper confidence level of 40.5%.

“Importantly, no 3D reconstruction was matched incorrectly and Picasa did recognize 28 of 29 reconstructions as a face,” Mazura said.

“Software reconstructions of CTs contain sufficient detail to allow matching of subjects’ faces in one out of three cases,” Mazura said. “This suggests that subjects whose head CT data resides in research repositories have the potential to be identified, even if protected textual data is removed. If accuracy rates can be improved significantly, additional de-identification will be required to distort the soft tissue components. Perhaps this is a new way to think about privacy.”

A member of the audience suggested the problem may be overstated. Things such as facial hair, glasses, and the use of profile images can make facial recognition software unusable, she said. If it’s one in three cases with perfect conditions, why should you care, because you rarely have perfect conditions?

Mazura said government agencies are imposing guidelines to make facial image displays more consistent.

“Think about passport or DMV databases,” he said. “I feel there are databases of these photographs that can potentially be used.”

Do you think stripping out textual identifying information in publicly available radiologic images will protect you against privacy violations? Think again. A paper presented Thursday at the 2010 RSNA meeting showed how facial images reconstructed from maxillofacial sinus and cerebral vasculature images could be matched in a database using commonly available face-matching software.

Reconstructions of radiologic facial images could strip away patient privacy

CT scout or scanogram images make up only about 4% of the typical chest/abdominal scan radiation dose, but are an easy target for dose reductions, according to a study presented Wednesday. Further, as technology changes and protocol updates reduce overall dose rates, scout images will make up a relatively larger part of the total and still represent a good target for cuts.

Researchers at the Baltimore VA surveyed radiologists on their use of scout scans and found they hold little value. They’re mainly used by technologists to assure the region of interest is covered in the diagnostic scan.

The survey of radiologists found that 63% almost never looked at the scout image, said Dr. Amee Patel, a resident, who presented the study. The survey also found a wide variation in actual dose levels used during the collection of scout images with one site 80% lower than others.

The most common setting has the scout image at 120 kVp/115 mAs, the equivalent of about 10 chest radiographs, and can represent up to 20% of the total CT scan dose, Patel said. Research has established that scout image dose reductions of 10-fold are possible while leaving overall landmarks still well visualized.

One problem encountered is that some scanners don’t permit lower settings for the scout images, Patel said.

Patel and Dr. Eliot Siegel, a coauthor of the study, both stressed the need to pay close attention to scout dose levels as lower dose imaging protocols emerge.

“In the National Lung Cancer Screening Trial, they are scanning at about 50 mAs. If the vendor would not allow them to go below 30 mAs, then the scanogram dose would be 50% of the entire scan, which is huge,” Siegel said. “We believe you go could all the way down to 3 mAs, which is similar to what we use for a typical test radiograph, and this would still allow the technologist to position the image with no problem. We believe you could reduce dose by 10- or 30-fold. The only reason we’re not is that vendors don’t allow us to go below 20 or 30 mAs.”

For cutting CT dose, start with the scout scan

It should be possible to reduce radiation dose levels by as much as 50% in CT appendicitis scans without seriously sacrificing accuracy, a study presented Tuesday by Duke University researchers concluded. 

Using a process that added simulated noise from reduced dose to actual abdominal images, the researchers compared the noisy images to the actuals and looked for reader agreement. They found that the simulated reductions of 25% and 50% were acceptable, but that a 75% reduction decreased confidence in the accuracy of the interpretation.

The study covered abdominal images from 100 patients. A noise introduction tool was used to create images with simulated drops of 25%, 50%, and 75% in the mA levels. Those images and the actuals were then randomized and read by three fellows and rated for diagnostic confidence, the ability to identify the appendix, and whether it was normal.

For all patients, agreement between readers on identification of the appendix on the original and the three reduced-dose scans was reasonably high. Means were 86%, 81%, 77%, and 77%, respectively. For patients without appendicitis, agreement on whether the appendix was normal was even higher; means were 91%, 91%, 90%, and 84%. For patients with appendicitis, mean agreement was highest at 100%, 97%, 100%, and 100%.

The researchers also evaluated the confidence levels of the readers, said Dr. Erik Paulson, who presented the study.

Using the actual images, most readers had confidence levels of 4 and 5, and even with a 25% dose reduction, the confidence level was “excellent,” Paulson said. When reducing dose to the 50% level, more 2 and 3 ratings emerged, and at 75% reduction, a loss in confidence was identified.

Paulson said there was excellent agreement around identification of the appendix at low dose, the presence of abnormalities, and the presence of disease, and that the accuracy was maintained up to a 50% dose reduction. He cautioned, however, that the simulation may overstate the actual amount of dose reduction.

“I think a reasonable take-home message is that even using a routine CT technique, reducing dose is warranted in patients with suspected appendicitis,” he said.

In the question-and-answer session, Paulson was asked how low dose could go in appendicitis scans.