What is a mammography?
A mammogram is an X-ray of
the breasts.
As part of the screening, it
is used to detect small cancers, well before they are palpable or symptoms appear.
It is carried out with a
specific radiology device called a mammogram.
One after the other, your
breasts are placed between two plates that squeeze and compress the breast for
a few seconds.
The doctor immediately
interprets the images and then carries out a clinical examination.
The clinical examination
involves palpating the breasts to detect certain anomalies that may not be
detectable on mammography.
An interview with the
doctor completes the examination.
Artificial intelligence goes further
Artificial intelligence
(AI), thanks to the emergence of deep learning and convolutional neural
networks, allows the development of cancer detection.
Also, it allows diagnostic
assistance systems with performances far superior to those of previously
available tools, which were not very specific.
The fields of application
of AI go even further, combining the sorting of images to be read in priority
by the radiologist for workload optimization.
That is including the
automatic identification of technically insufficient images, the standardized
assessment of breast density, and individual breast cancer risk.
Also, the AI offers the
reduction of Tomosynthesis slice reading time.
Nevertheless, although
multiple retrospective evaluations show very promising results, there is still
a lack of prospective studies in real screening conditions.
Artificial intelligence in mammography
Artificial intelligence. (AI) can improve radiologists' performance in reading breast cancer screening
mammograms, according to a study published in Radiology: Artificial
Intelligence.
Such software received US
FDA validation in March 2020.
In breast cancer screening
mammography, many malignant lesions go undetected and suspicious findings often
turn out to be benign.
According to an earlier
study in the journal Radiology.:
On average only 10% of
women were recalled for further diagnostic workup based on suspicious findings
that ultimately indicated cancer.
AI algorithm: better cancer detection
AI algorithms represent a promising solution for improving the accuracy of digital mammography.
Developers
"train" the AI on existing images, teaching it to recognize
abnormalities associated with cancer and to distinguish them from benign
findings.
The artificial
intelligence programs can then be tested on different sets of images.
AI not only offers the
possibility of better cancer detection, but also improved efficiency for
radiologists.
In a new study,
researchers used MammoScreen, an AI tool that can be applied with mammography
to aid cancer detection.
The new study was
published in Radiology, under the title of Artificial Intelligence.
The system is designed to
identify regions suspected of having breast cancer on 2D digital mammograms and
assess their probability of malignancy.
Fourteen radiologists
evaluated a dataset of .240. 2D digital mammography images acquired between 2013
and 2016 that included different types of abnormalities.
Increased sensitivity-shorter diagnostic time
The average cancer
sensitivity increased slightly when using AI management.
AI also helped reduce the
rate of false negatives, or results that appear normal even though cancer is
present.
"The results show
that MammoScreen can help improve radiologists' performance in detecting breast
cancer," said Dr. Serena Pacilè.
Dr. Serena Pacilè is the
head of clinical research at Therapixel, where the software was developed.
The improved diagnostic
performance of radiologists in detecting breast cancer was achieved without
extending their workflow.
In cases with a low
probability of malignancy, the reading time was reduced in the second reading
session.
This reduced reading time
could increase radiologists' overall efficiency, allowing them to focus their
attention on the most suspicious examinations, the researchers said.
MammoScreen validation by US FDA
In March 2020, the US Food
and Drug Administration approved MammoScreen for use in the clinic, where it
could help reduce radiologists' workload, according to Dr. Pacilè.
The researchers plan to
explore the Artificial Intelligence tool's behavior in a large screening-based
population and its ability to detect breast cancer earlier.
