Debbie Rowe, who religiously
performs a breast self-exam every month, is having her
first-ever mammogram. Her mother had breast cancer at
35, Debbie's age, and she's worried that she may easily
miss a lump herself. Mammograms can detect breast cancer
early based on the appearance of tiny calcium deposits
or visualization of a minuscule tumour. However, the technology
can produce false negatives � sounding the all-clear when
trouble is brewing. This may occur less frequently in
future, thanks to a report in the August issue of Radiology,
which described how a computer-aided visual boost dramatically
increased the ability to detect the potentially cancerous
hotspots.
The study was conducted at The
Elizabeth Wende Breast Clinic in Rochester, New York,
under the direction of Dr Stamatia Destounis. The researchers
twinned a commercially available computer-aided detection
(CAD) software with their x-ray imaging system. "The
CAD system software scans and digitizes the films, marking
microcalcifications and masses with triangles, which
most of the time are benign but once in a while they
can be very early ductal carcinoma in situ," Dr Destounis
explained.
"Our study shows that the CAD system
can increase our detection rate of cancer to less than
or up to a year earlier than what we are currently diagnosing
it on screening mammography," she added. The evidence
for this dramatic conclusion comes from the retrospective
study conducted in 2000 that involved 519 histologically
diagnosed breast cancers. Of these, 318 patients had
had a mammogram sometime prior to the diagnosis.
The 318 x-ray films were re-analyzed
by five radiologists who were in the dark as to their
colleagues' interpretations of the images. Between them
they spotted 98 false-negatives, of which 71 represented
visible cancers. At least three of the five radiologists
deemed 52 of the visible cancers to be "actionable."
The CAD system was used to analyze
the x-ray films of the 52 critical images. Thirty-seven
of these were correctly pegged as being images of cancerous
growth. In the first go-round several years before,
the two radiologists who had evaluated the same images
had determined that these were negative. "It appears
that CAD has the potential to decrease the [false-negative]
rate at double reading by more than one third (from
31% to 19%)," concluded the researchers.
Does this mean that breast cancer
diagnosis has entered an I, Robot era, where
it can all be done remotely? Certainly not, according
to Dr Destounis. Indeed, for every one CAD-assigned
mark that was correctly identified as cancer there were
two marks that were false alarms. "This system will
never replace the radiologist. It just marks something
for us to take a look at," Dr Destounis stated.
Increased accuracy of screening
mammograms will be good news to the estimated 21,200
Canadian women who will be diagnosed with breast cancer
this year.
|
