is the Clinical Section Head of Imaging Informatics, Geisinger
Health System, Danville, PA.He is also a member of the editorial
board of this journal.
I am sitting in a mountain cabin on the Sunday after the RSNA
annual meeting, recovering from another week of social and
professional hyperstimulation. Snow from an early December storm
covers the skylights, creating a subdued light that, sadly, is
reminiscent of the PACS reading room to which I will be returning
tomorrow. My reverie shifts to anxious anticipation as the
peacefulness of the afternoon gives way to thoughts of the chaos of
the clinical fray. A quick check of the online schedule--an
innovation I have, until recently, happily done without for 20
years--tells me my assignment. I will be covering mammography
tomorrow. I wonder how many days behind we are. Surely after the
costly and time-consuming installation of full-field digital
mammography in July, we must be better off than last year. Not so.
The backlog is greater than ever, as radiologist efficiency has
suffered in the stampede to soft-copy mammographic interpretation.
Were we right to turn to this immature technology that has yet to
demonstrate clinical benefit for all patient cohorts? Perhaps we
made a tactical error in making the switch at this time.
As a recent shell-shocked refugee from a small private practice,
many ask my opinion of the contrast with my current, more academic
position in a much larger department. Small private practice, I
reply, is dog-eat-dog, but in academics, it is the other way
around. The timeless constant is the endless assembly-line supply
of images waiting to be manufactured into a radiologic diagnosis.
Radiologists have, to some extent, become the 21st century
pieceworkers; we have supplanted the sartorial laborers of the
Industrial Revolution, who were paid by how fast they could attach
buttons and lace in dangerous and uncomfortable sweatshops. Some
laborers of this era worked at home--eerily foreshadowing today's
grid-working teleradiologists. We must be crazy to emulate this
shameful period in history.
Just what is my specific beef with digital mammography? One
word: workﬂow. Most, but not all, radiologists are
reporting longer interpretation times for digital mammography
versus film-screen mammography. In many ways, reading protocols are
more sophisticated than their PACS cousins. However, they are often
fully operable for only a specific vendor brand. In many
enterprises, multiple vendors are used across a system because of
complex political or economic reasons. Even if a single vendor is
in use, outside studies for comparison or second interpretation are
often not adequately accommodated.
Worklist functionality has also not kept pace. Digital
mammography vendors seem to have learned little from the
conventional PACS experience, even when the same manufacturer
produces both systems. Better accommodations for multiple
concurrent readers and more sophisticated filtering functions are
Conventional PACS vendors are still designing their software
around the mouse and the hopelessly parachronistic computer
keyboard. At least mammography vendors are taking a step forward in
using an alternate user interface device. Yet this technology could
also be improved with the inclusion of more user configurability.
Not all navigation commands are currently available on these
devices, which still require the distraction of a pulldown menu.
This imperfect user interface remains in direct opposition to our
primary goal of detecting subtle findings with minimal distraction.
"Eyes-free image interpretation" should be the rallying cry of
I reserve my most sulfurous resentment, however, for some of the
specialty BI-RADS report-creation software. These have not kept
pace with the current state-of-the-art reporting systems. With the
decreased efficiency of soft-copy reading, it is even more
important to minimize extraneous tasks at the workstation, and the
majority of these relate to creating a report. The look-away time
that some reporting systems require is simply unacceptable.
Point-and-click technology must give way to hybrid reporting that
combines speech recognition with the structure necessary for data
tracking and patient notification. Moreover, this software must be
fully integrated with the worklist and navigation features of
What was that? I'm certain I heard a mufﬂed crash
through the drifting snow. A tree falling in the woods? No, it
sounded more like one of our vendor advertisers hitting the
ﬂoor. At the risk of being labeled a plodding Luddite, I
should expand these thoughts.
No, we did not err in moving to this technology. After all, PACS
in its infancy arguably suffered from similar detractions. The
current ﬂaws in digital mammography workstations are not
permanent; they are merely a self-limited case of adolescent acne.
With guidance from IHE profiles, vendors will become more
interoperable. Conventional PACS workstations are rapidly becoming
suitable for digital mammography. Stereotactic biopsies and needle
placements are now much faster, improving patient comfort as well
as radiologist efficiency. Ultimately, integration with CAD
findings should allow the automated creation of a full report
needing only a single radiologist command for report approval and
sign-off. In less than 2 years, we will have a markedly decreased
dependence on old films or digitized prior studies, streamlining
the now hybrid comparison process. Tissue equalization already
makes subtle lesion detection in dense breast tissue easier.
Tomosynthesis and stereoscopic viewing should further increase
diagnostic accuracy. No, this is not an immature technology, but
rather a simple case of taking one step back before two steps