Google AI Algorithm Offers Better Results in Breast Cancer Diagnosis
Mammograms play a crucial role in breast cancer detection. They are the most common tool which experts use for diagnosing cancer, but they are not a seamless screening tool. Sometimes, experts miss tracking signs of cancer, which might lead to a more severe form of the disease. Now, a discovery of researchers from the US and UK could make the process more accurate. A new AI program is offering more efficient results at cancer diagnosis using mammograms. The study, published in journal Nature, performs better than radiologists and accelerates the existing processes of diagnostic. Currently, radiologists are unable to detect breast cancer in one among five patients through traditional mammogram screening. As per the American Cancer Society, experts overlook around 20% of breast cancers in this testing.
The latest computer program has gained support from Google’s DeepMind artificial intelligence unit. Also, it includes researchers from Britain’s National Health Service and the Imperial College of London. The team collectively has trained the algorithm to detect breast cancer on over 28,000 mammograms or say X-ray images. Even more, they have equated the computer’s outcomes with the actual reports of patients. The study has estimated two massive sets of mammograms, one from the US and the other from the UK. The UK dataset includes mammograms of over 25,000 women, and another consists of more than 3,000 females from the US.
In the US collection, the program has produced a 5.7% less false-positive diagnosis. It is better than radiologists or human experts. Apart from this, the computer program has recorded 9.4% less false negatives, signifying it has selected various breast cancers overlooked by radiologists. Above all, if the AI technique reveals its worth in clinical researches, Google software could make the screening process more powerful. Even more, it could lighten the burden on health services like the NHS, where there is a shortage of radiologists. Dominic King, a UK health researcher working at Google Health, said it is a great example of how these technologies can empower and expand the expertise. Even more, Google expects that the system can be ultimately used in medical practice.