A new study suggests that a Google AI system could better identify breast cancer than specialists in radiology. Compared to medical professionals, the research saw the computer created by Google’s AI experts as both screened mammograms. AI has been shown to be more effective than humans in predicting breast cancer cases and it is far better to avoid false-positive results.
Similar technology will now be used by health experts to improve other health issues along with breast cancer detection rates which affect one out of eight females. Scientists from the US and Great Britain conducted the study and published it in the “Nature” journal. The latest proposal suggests that AI will lead to drastic health-care improvements.
The American Cancer Society reports radiologists to skip approximately 20 percent of mammograms of breasts, and a false positive outcome is available among half of all women tested over a 10-year period.
Mozziyar Etemadi, one of the co-authors from Northwestern Health in Chicago, said:
The findings of the study, developed with Alphabet Inc’s DeepMind AI unit, which merged with Google Health in September, represent a major advance in the potential for the early detection of breast cancer.
The team, which included researchers from Imperial College London and the NHS, conducted the program on thousands of mammographs to classify breast cancers. They then compared the performance of the system with the actual results of a set of 25,856 mammograms in the UK and 3,097 from the US.
The study showed that the AI system was able to identify cancers with a similar degree of accuracy to expert radiologists while reducing the number of false-positive results by 5.7% in the US group and by 1.2% in the British group.It also reduced the number of false negatives, where results are incorrectly marked as normal, by 9.4% in the US group and 2.7% in the British group.
Such variations reflect the reading ways of mammograms. In the United States, only one radiologist reads the findings and examinations are done each year or two. Tests are performed in the UK every three years, and two radiologists interpret. A third Dr. is contacted if they disagree.The group tested six radiologists against the AI system and found that they were superior to the effective diagnosis of breast cancer. Connie Lehman, Head of the department of breast imaging at Harvard’s Massachusetts General Hospital, said that:
The results are in line with findings from several groups using AI to improve cancer detection in mammograms, including her own work.
For mammography laboratories, computer systems are often used to increase the diagnosis of cancer. CAD technologies for nursing settings have not been able to improve clinical results.
The problem, Dr. Lehman said, is that existing CAD systems have been trained to identify conditions that can be seen by human radiologists, while machines learn to detect cancers with AI, based on the actual results of thousands of mammograms.
The research is somewhat minimal. Most of the studies were carried out using the same form of imaging equipment, and a lot of patients were with confirmed breast cancers in the US community.
Crucially, the team has yet to show the tool improves patient care, Dr. Lisa Watanabe, CureMetrix’s chief medical officer, whose AI mammogram system received last year’s approval from the US.
“AI software is only helpful if it actually moves the dial for the radiologist,” she said. Dr. Etemadi accepted that such studies, including regulatory approval, are required, a process that could take several years.