A few years ago Artificial Intelligence seemed like a science fiction scenario. Today, its possibilities are not limited to theoretical discussions between specialized scientists but find applications in our everyday life. The case of the tool is illustrative MIA which pilot analyzed more than 10,000 mammograms of women in a UK NHS hospital.
The Artificial Intelligence tool correctly diagnosed 90% of patients which the doctors had also identified, a percentage that is considered particularly high. In addition, it identified an additional 11 women with cancerous tumors that doctors were unable to see with the naked eye due to their extremely small size. Which means that the collaboration of humans and artificial intelligence “gave birth” to the most effective diagnostic tool for breast cancer.
Gerald Lipp.
“All 11 women are now in good health, they underwent operations and needed only a few days of radiotherapy. Artificial intelligence therefore proved to be a lifesaver for them, as it detected the tumors in time. Without this tool, these women would have gone on with their lives as normal, thinking they were healthy according to the doctors, but on re-examination the tumors would have grown and possibly spread. This, in turn, would lead to more serious surgeries and heavier radiotherapy and, perhaps, chemotherapy” he tells BIMA Science o Gerald Lippclinical director of breast screening in North East Scotland and the doctor who led the project.
Dear cooperation
The decisive result for the lives of these women raises the so far unanswered question: “Will machines eventually replace humans and for that matter doctors?”. Dr. Lipp answers emphatically no. “In Britain mammograms are always read by two radiologists. The former detects most cancers and the latter can detect cancers that the former did not detect. AI works in a complementary way, as it can find abnormalities very small in size that humans cannot».
2.3 million women worldwide developed breast cancer in 2022 and 670,000 died, according to data from the World Health Organization.
In addition, he says, the MIA machine has not been trained to “read” the genetic characteristics of different populations, and this carries the risk of false positives. “In multicultural areas such as London, machines will need to be properly trained. Because black populations, for example, may have increased breast density and this may be perceived by the machine as suspicious. Also, the machines do not currently have access to the history of each patient and something benign that was previously detected by doctors and did not inspire concern to be read as a malignant tumor. Therefore, radiologists are more than necessary despite the development of technology,” he explains.
374,800 women were diagnosed with breast cancer in Europe in 2022 and 95,800 died.
Still, he says, it’s a valuable tool in doctors’ quiver, saving time, allowing them to focus on more complex cases. “There are days when we are asked to read 100 mammograms in a few hours. When a radiologist diagnoses a mammogram as negative and the machine gives the same result, we proceed to the next one, no need for a reading by a second doctor. During the pilot implementation of the program, we found that the amount of work involved in reading the scans was reduced by 30%.”
In the service of diagnosis
According to Dr. Gerald Lipp, the future of artificial intelligence in diagnostics is very promising, as specially trained machines will be able to predict the occurrence of cancer in the next five to ten years. “We’re still at the beginning, kind of like Windows 3.0 in the 90s. In the future, with the help of this technology, we will be able to accurately calculate the chances of breast cancer and indicate the necessary frequency of re-checks.
»Artificial Intelligence is already being applied to reading biopsies with impressive results, while soon it is expected to be used in lung and prostate screening tests as well as in CT scans of the brain. This will help detect any very small lesions that doctors were unable to find, as was the case with the 11 women. At the same time, the time required for an MRI will be significantly reduced,” he says, pointing out that along with saving time, the stress of the patients is also reduced.
29.4% of cancers detected in women in Europe in 2022 were breast cancer.
“Besides the fact that each clinic will have the possibility to carry out more MRIs in one day, we must not overlook the positive impact on the anxiety of waiting for patients and also on that of claustrophobia” he adds. What for him is of particular importance for the improvement of prevention and care services in general is the commitment of experts to the continuous upgrading of artificial intelligence applications and the education of women.
In which countries does it apply?
In the United Kingdom tests of mammogram analysis with the help of artificial intelligence continue in various public hospitals and after this technology is approved by the National Preventive Medicine Council it will be widely applied and provided free of charge by the British Health System. This, according to Gerald Lipp, is particularly important because no one will be exempt and consequently there is no question of inequalities.
On the contrary, on USA the test is not covered by health care plans, and women who want to have it will have to pay. It is worth noting that some time ago the tests of artificial intelligence in reading mammograms were completed in Denmark and the technology is already being implemented in Copenhagen and the region of Southern Denmark. Finally, at Sweden the “smart machines” are used both in symptomatic patients and in the context of breast cancer screening at no additional cost to patients.
#Artificial #Intelligence #breast #cancer
