Current state-of-the-art artificial intelligence technology or improved detection of breast cancer

deep-learning artificial intelligence, widely used to detect targets
in images, can improve early breast cancer detection, according to a study published in Scientific Reports.
deep learning is a branch of machine learning. Its network is inspired by knowledge of how the biological brain, such as the brains of humans or animals, works. Deep learning networks "learn" from human-annotated data set to achieve image recognition capabilities similar to those of humans.
based on state-of-the-art deep learning, Dezso Ribli and colleagues have proposed an improved computer-aided detection (CAD) system that can be trained for the detection and positioning of breast lesions. For the past 20 years, the CAD system has been used to assist radiologists in detecting breast cancer: analyzing mammograms, marking suspicious areas and then reviewing them by radiologists. However, the cost of using these technologies is expensive ($400m a year in the US) and the results are still controversial.
When the authors tested using a database of mammograms of 115 confirmed cancer patients (2 out of 4 images in each case), the authors' improved CAD system identified 90 percent of malignant lesions without human intervention, with very few false positives. The current screening method, which includes radiologists, has a 77-87% correct rate for cancer identification.
these findings mean that expensive traditional CAD methods could be replaced by cheaper methods of deep learning, which are now used to identify objects in traditional images, such as dogs and cats, to help radiologists detect more cancers. So far, however, the authors note, they have tested their methods only on small data sets of confirmed cancer images. (Source: Science.com)