They present a new AI system that reduces false positives in the detection of breast cancer

1634836389 cáncer de mama.jpeg
1634836389 cáncer de mama.jpeg

Within the wide potential of action that artificial intelligence has in the scientific world, we periodically learn of new efforts to use this technology as an instrument that enhances the work carried out in various fields. Among those, medicine stands out as one of the main ones, which among other challenges has been targeting cancer for a long time.

Under this point, recent advances have emerged that, with the help of AI, focus on the timely and adequate detection of breast cancer, such as a system developed by Google that analyzes data from mammograms and another mechanism that originated in the University of Waterloo, which analyzes body-friendly microwave data. To the list we add a new system recently introduced, which promises to specify the diagnoses made when analyzing ultrasound examinations.

Perfecting imprecise technology with the help of artificial intelligence

One of the most used mechanisms for the detection of breast cancer is through localized radiographs, popularly known as mammograms. In parallel, as an alternative, ultrasound analysis is present, a technique used in ultrasound.

The use of ultrasound is less expensive than mammography examination. Apart from the economic advantage, this technology stands out for not exposing its patients to radiation. However, the main deficiency that keeps this technology away from massive use in laboratories is its wide margin of error, yielding a higher than average number of false positives associated with a conventional mammogram.

This innovation, which promises to enhance the weakness of ultrasound in this field, was developed by researchers at New York University. The details of the development presented, mostly technical, were published by the magazine Nature Communications. Among the observations shared there, it is pointed out that the new ultrasound system that they presented increased its precision from 92% to 96% and, in addition, it can reduce the performance of unnecessary biopsies.

Linda Moy, a New York University radiologist and part of the research team, commented: “If our efforts to use machine learning as a classification tool for ultrasound studies are successful, ultrasound could become a more effective tool in the detection of breast cancer, especially as an alternative to mammography and for those with dense breast tissue ”, adding that this development could have a strong impact in your area.

The researchers will continue to work on refining this artificial intelligence tool. The following steps will consider new risk factors in their formula, such as genetic predisposition to this type of cancer.