Artificial intelligence for early cancer detection. What would you think if I told you that Artificial Intelligence can be used for early cancer detection?
Google has collaborated with partners such as DeepMind and institutions such as Northwestern University, the Cancer Research UK Imperial Center and the Royal Surrey County Hospital, among others, to investigate the possibilities of applying artificial intelligence to breast cancer detection.
As in other medical disciplines, they have discovered that an AI model can be developed that can thoroughly examine diagnostic imaging tests, such as mammography, to increase the effectiveness of breast cancer detection.
According to the results just announced by Google, and which appear in an article published in the journal Nature, the system they propose has reduced false positives by 1.2% in the cases studied in the United Kingdom, and 5.7% in the Americans. It has also been able to reduce false negatives by 2.7% in the UK, and by 9.4% in the US. In addition, those responsible for this research have used the data obtained in the analysis of the cases in one country, and applied it to the cases in the other, reducing the percentage of false positives by 3.5% and false negatives by 8.1%.
This clearly shows that, even in a preliminary investigation, AI has enormous potential to improve the work of analyzing visual information in diagnostic tests. Based on these results and on the ability that AI has demonstrated in the analysis of other data such as those from analyses and other types of tests, experts say that much more accurate models can be developed to improve diagnosis.
A recent study conducted by a team of neurosurgeons at the University of Michigan details how AI technology was successfully used to diagnose brain tumors during surgery, while patients were still on the operating table.
“AI is invaluable in the operating room,” said Daniel Orringer, M.D., one of the neurosurgeons who co-authored the study. “It is often impossible to make the distinction between tissue containing a tumor and tumor-free tissue with one eye, and even this can be difficult to achieve on a microscopic scale.
Usually, after surgery, it takes a pathologist about 30 minutes to study the results of a biopsy under a microscope and make a diagnosis. During that time, the patient is sewn up and sent to the recovery room to wait for the results.
With the IA the results were achieved in an average of two and a half minutes. This allowed the surgeons to understand the complete medical picture of the patient, and determine what the best care was, while everyone was still in the operating room.
Orringer explained that different tumors are treated differently; sometimes it is better to treat them with an operation and others respond better to chemotherapy and radiation.
“Being able to match the tumor with the right treatment is essential to patient care,” he said.
In addition, AI and human pathologists achieved roughly the same number of diagnoses in 278 patients under study. Using AI alone, the cancer diagnosis rate was 94.6 percent, while pathologists achieved 93.9 percent.
“We are very optimistic about the chances of AI to diagnose tumors,” said Orringer. He then added that “better diagnosis means better care. Better care allows for the possibility of more effective treatments for cancer.
And, moving further into the future, the researchers’ goal is to improve early detection systems through AI of this type. This would mean making a very important qualitative leap in the field of preventive medicine, being able to treat patients before the disease becomes apparent and in a totally personalized way, taking into account the accumulated experience but applying this knowledge and capabilities in an individualized way.
¡Stay tune! Go and visit our blog. New articles every week.