Survival of the Slickest

“Radiologists are the imaging experts …” or are we? Are we the experts because we say so, or is it because of our training? Completing a fellowship program in radiology, or for that matter a general diagnostic residency, in and of itself, does not impart expertise. He/she who does it best should do it! If the specialist does it better, they should be allowed to do it.

As an example, an orthopedist may have greater depth in understanding orthopedic anatomy, pathology, and physiology than a radiologist does. To perform imaging, such a specialist would have to have an understanding of, and indeed be certified in imaging safety, as do all of us that are certified in our specialty. However, Stark was developed to obstruct and outlaw fraud, abuse, and self-referral. Therefore, I suggest that we allow specialists to perform imaging with the following provisos: 1) they need to be certified in imaging safety and 2) they cannot self-refer; that is, they cannot earn revenue/income both as clinicians who examine patients and as referrers who refer to themselves for imaging (I have avoided discussing the conflict that may also arise from self-referral leading to surgery or treatment that may arise from the imaging interpretation).

In principle, Stark was formulated to prevent just thisself-referral! If an orthopedist or any specialist feels that they can perform imaging better than their local radiologist (and perhaps they can), they must become imagers, give up their clinical practices, and compete with radiologists on a level playing field. If they compete successfully by these rules, they have the same opportunity to enjoy the practice of imaging and to earn their living in this way. When they invest millions of dollars in imaging equipment, they must take the same risk that we as radiologists take, and they must compete based on quality and service. They cannot be guaranteed sufficient volumes through self-referral to make their lease payments.
Irwin A. Keller, MD, neuroradiologist at the University Radiology Group, PC, East Brunswick, NJ

MDCT: A Disruptive Technology

I would like to pick a nit concerning the article in the October Axis Imaging News entitled “MDCT: A Disruptive Technology Evolves.” The term “disruptive technology” has a specific meaning, as defined by Clay Christensen. MDCT is about as opposite as you can get from a disruptive technology. It is a poster child for a “sustaining technology.” I would suggest reviewing “The Innovator’s Dilemma” for a discussion of the two terms. I firmly believe that radiology is susceptible to disruption in the classical sense as described by Christensen (if something that was described less than 10 years ago can be considered classical), but MDCT and other high-end technology will not be the agents.
Lawrence A. Liebscher, MD, Allen Imaging Center, Waterloo, Iowa

EDITOR’S NOTE : It should be noted that the headline was chosen by the editor, not the author. A better choice would have been, “MDCT: A Sustaining Technology Disrupts.”

Data Flow and Infrastructure

I was reading West Long’s paper entitled “Data Flow and Network Infrastructure” in the October 2004 Axis Imaging News supplement, and unfortunately, I could not help but notice a recurring error in Long’s estimation of image size.

Long should know that cardiology produces primarily dynamic images (multi-frame) according to the obvious dynamic nature of the heart. If we use the numbers Long is referring to, a modality producing images 1k x 1k, 12 bits, will generate 100 to 200 times the amount of data Long is suggesting, making CT and MRI study size secondary compared to cardio studies. DICOM XA multi-frame images should be taken into account at all times since still frame size is so small it becomes irrelevant compared to multi-frame.

For instance, a study composed of 15 recorded cine-loops each around an average of 5 seconds long (according to our average study length) will generate: 15 loops x 5 seconds x 30 images x 1024 pixels x 1024 pixels x 2 bytes/pixel = 4.719 MB/study

This equation does not take into account a lossless compression (ratio around 2:1) that is generally applied. The size of the study could also be cut by 2 if you were using 15 frames per second which could be acceptable for some institutions. Taking both into account a compression of 2:1 and a reduced frame rate, we will still generate 1 GB studies in average. This is still 30 times larger than the size Long suggests.

I think this omission should be highlighted since it’s present in every table of Long’s paper.
Stéphane Morin, PEng, MScA, Biomedical engineering, Hopital Laval, Heart and Lung Institute, Quebec, Canada