Computer-aided detection has the potential to improve the practice of screening mammography significantly. This article resents the results of a study of two suburban, general radiology practices designed to examine the impact of CAD on the workload of radiologists and technologists.
Dr. Shile
is Medical Director of the Susan G. Komen Breast Center,
Peoria, IL.
Computer-aided detection (CAD) has the potential to improve the
practice of screening mammography significantly. While current
clinical trials will help define how this technology will enhance
the diagnostic efficacy of screening mammography, there is little
information on how CAD will affect workload in radiology practices.
A study was performed of two suburban, general radiology practices
to examine the impact of CAD on the workload of the radiologists
and technologists.
Materials and methods
Practice sites--Both radiology practices offer breast imaging
services and are staffed by radiologists, who subspecialize in
breast imaging, and dedicated breast imaging technologists.
Screening mammograms are batch read by the interpreting
radiologists several times a day.
CAD technology--The study examined the use of the Second Look*
(CADx, Laval, Quebec, Canada) computer-aided detection system. Exam
films are batch loaded into the system's digitizer for computer
processing. The system's output, known as a Mammagraph, is a paper
printout of the screening exam with potentially suspicious masses
and microcalcifications identified in the exam images. This output
serves as a reading aide to the interpreting radiologist and can be
thought of as an independent second opinion to be consulted during
exam interpretation.
Data collection--Data was collected in order to identify the
amount of additional time required by technical staff and
radiologists to process and interpret exams as a result of their
use of this CAD technology. Timing data were collected using
videocameras mounted in technologists' processing areas (where the
Second Look devices were located) and screening mammography reading
areas. Videotapes were analyzed by a Mammography Quality Standards
Act and board-certified radiologist.
Radiologists--Under institutional review board approved
protocols, two radiologists were studied during their
interpretation of screening mammograms. They both interpreted and
reported on 35 consecutive exams from their practices. After
completion of their usual exam interpretation and report, they
reviewed each exam with the Mammagraph and modified their
interpretation and report as needed. For each exam, the total time
for interpretation without the aid of the Mammagraph was
determined. This included the amount of time spent assessing the
exam with a magnification lens, reporting findings,
interpretations, and recommendations. The amount of time added to
case interpretation as a result of the radiologist's review of the
Mammagraph was also determined.
Technologists--Technologists at both study sites were studied
while processing cases. Timed events included removing exam films
from X-ray film jackets, arranging and identifying films for Second
Look processing, loading cases into the film digitizer, and
refiling exam films and associated output into x-ray film jackets.
The total amount of technologist time spent on these tasks was
determined for 35 screening mammograms at each study site.
Data analysis--Mean and standard deviation, median, minimum, and
maximum process times were calculated for the tasks performed by
the radiologists and technologists.
Results
Radiologists--Table 1 lists key events during the radiologists'
interpretation of exams without the use of the Mammagraph output.
The mean time for case interpretation was 1 minute and 16 seconds,
which included 16 seconds of assessment with a magnification lens
and 16 seconds of case reporting. On average, the use of Second
Look technology required an additional 17 seconds per case of a
radiologist's time, with a minimum of 3 seconds per case and a
maximum of 38 seconds per case (Table 2).
Technologists--On average, an additional 1 minute and 21 seconds
of technologist's time was required for case processing by Second
Look (Table 3).
Discussion
This study was performed to assess incremental changes in
workloads that occur with the use of Second Look CAD technology in
screening mammography. For radiologists, average case
interpretation time increased from 1 minute, 16 seconds to 1
minute, 33 seconds; an additional 17 seconds. Compared to other
events that occur during exam interpretation, it is interesting to
note that the work effort created by the use of Second Look is
roughly equivalent to that required to assess an exam with a
magnification lens, 16 seconds. If its diagnostic efficacy is
equally great, one can speculate that CAD will be as essential in
mammography interpretation as the magnification lens is today.
Clinical trials currently in progress should help to address this
issue.
The additional amount of time needed by technologists to
implement Second Look CAD technology was 1 minute, 21 seconds.
Thus, in facilities where screening mammograms are scheduled at 15
to 20 minute intervals, the findings suggest that technologists can
implement the technology as they perform patient exams without any
change in patient throughput.
Conclusion
This study was designed to determine the additional amount of
time that is required by radiologists and technologists to use
computer-aided detection in screening mammography. On average, CAD
adds an additional 17 seconds of radiologist's time to the
interpretation of a screening exam. This is about the same amount
of time that the radiologists in this study used the magnifying
lens during exam review. CAD also increases technologist's case
processing time by an average of 1 minute and 21 seconds per exam.
AR
* Second Look is for investigational use only and is currently
not for sale in the United States.