How AI can detect breast cancer, according to a new study

How AI can detect breast cancer, according to a new study

Almost daily, new scientific research demonstrates the importance of artificial intelligence in the medicine, as it allows not only to analyze large-scale data, but also to detect with Greater accuracy and velocity different pathologies. In this regard, research by Swedish scientists warned that the AI can still be more effective than traditional methods of reading mammograms.

While noting an additional advantage: Reduces workload for radiologists by 44.3%.

A study published in The Lancet, conducted at four screening sites in Sweden on more than 80,033 women aged between 40 and 80, warned that AI identified a 20% more cancers compared to standard human expert review.

“Retrospective studies have shown promising results using artificial intelligence (AI) to improve mammogram accuracy and reduce screen-reading workload; however, as far as we know, a randomized trial has not yet been conducted,” the scientists from Lund University in Malmö, Sweden, said in the paper. While noting that their goal was to analyze “the clinical safety of an AI-compatible screen reading protocol.”

As they explained, it was a randomized controlled trial, in which participants were randomly assigned to groups that received AI-assisted screening or non-AI double standard reading.

The results showed that AI screening identified 244 cancers compared to 203 detected in the control group. False positive rates were similar in both groups.

“The Positive Predictive Value (PPV) of recall was 28.3% in the intervention group and 24.8% in the control group,” they said in the paper. They added: “Screen reading workload was reduced by 44.3% with AI.”

In that sense, according to the study, the cancer detection rate in the intervention group with AI was approximately 0.61%, while in the control group without AI it was approximately 0.51%. This represents a difference of 0.10% in favor of the group with AI. That is, detection in the intervention group was approximately 20% higher.

This study enhances the information that, so far, has been evidenced on the role of AI in medicine. “The artificial intelligence and machine learning today are used in the oncology, and help Speed up your search of new information and in ordering it,” he had explained. Diego Kaen (MP 1898), president of the Argentine Association of Oncology (AAOC), to Infobae on a recent note.

At the same time as the doctor and researcher Daniel De Florian, leader of the International Centre for Advanced Studies of the School of Science and Technology of the University of Sint Maarten (UNSAM), he said: “What this does technology is learn from certain previous patterns in order to anticipate future behaviors in different circumstances and in any system.” “That ability to learn has to do with previous experiences. That is why it is not an innate intelligence but it is acquiring and becoming more complex. In that sense it is like the human mind.” he had said to Infobae About the operation of artificial intelligence.

“The artificial intelligence is Revolutionizing everything we know, including the medicine, but is still incipient“, he had stated to Infobae researcher Ezequiel Álvarez, De Florian’s colleague at UNSAM’s International Center for Advanced Study.

“In many ways these machines perform better than the human being. If we handle ten variants, the machines handle a thousand without any effort. But the final decision and what is done with that is up to us,” he added.

In that sense, Kaen added: “Today there are certain specialties within cancer treatments, such as diagnostic imaging, where Artificial intelligence helps to find alterations in CT scans, in mammograms. In these areas, where doubts were sometimes generated, artificial intelligence helps to unify everything and to begin to divide benign pathologies and malignant pathologies.”

“The human genome project was the kick-off for what is now precision medicine. It is to understand that tumors have their ID or have their genetic characteristic that is different from one patient to another. Today, we can also study the tumor genome and from there see what expression it generates, what mechanisms are called driver or behavioral and what is producing carcinogenesis, that is, it is giving a signal for a cell to reproduce in an unbraced, anomalous way without brake, “said Kaen.

He added: “This allows us to study what are those drivers or genes that produce conduction of the carcinogenesis signal, would be, and to be able to inhibit them to inhibit that carcinogenesis. And this brings a clear clinical benefit.”

Whereas Ezekiel Alvarez, he concluded: “We already have the knowledge so that the artificial intelligence do everything; It has already been understood how it works and Every month studies will appear“.


Please enter your comment!
Please enter your name here