Google AI can diagnose breast cancer better than doctors, but its too early to replace it

category:Internet
 Google AI can diagnose breast cancer better than doctors, but its too early to replace it


Unlike humans, artificial intelligence is indefatigable. Experts say this can improve detection rates and speed up diagnosis - computers can analyze images and give results in seconds. Our team is proud of these findings, which shows that the tools we are developing can help clinicians find breast cancer more accurately, said Dominic king, an expert on Google health

The researchers also said artificial intelligence could reduce the need to invite two doctors to judge mammograms and relieve doctors work pressure in the event of a shortage of radiologists. Allah Darzi, co-author of the report and director of the Empire Centre at the British Cancer Research Centre (CRUK), told the BBC: this is far beyond my expectation. It will have a significant impact on improving the quality of diagnosis, and it will also give radiologists the freedom to do more important things.

Most of the mammograms that AI judges come from Cancer Research UK databases, including 25000 images from hospitals across the UK, and more than 3000 from Northwestern Memorial Hospital in Chicago, us.

The results showed that compared with radiologists who initially saw these images and gave diagnosis, AI reduced 9.4% and 2.7% of missed cases in the United States and the United Kingdom, respectively. Misdiagnosis of breast cancer was also reduced by 5.7% and 1.2%, respectively. In addition, the research team invited six American radiologists to diagnose 500 randomly selected images, compared with the performance of artificial intelligence.

But in some cases, these radiologists can find cancer missed by AI.

The real world is more complex and possibly more diverse than the type of research environment the study is restricted to, ETA Pisano, chief research officer of the American Society for radiology, commented in nature

Pisano explained that the study means that artificial intelligence could one day assist in the early diagnosis of breast cancer, but the study did not cover all existing mammography systems. Apart from age stratification, the study has no other indicators to represent the general population, that is, it is not widely applicable.

Breast cancer is the highest incidence of malignant tumor in Chinese women, with the incidence increasing year by year. In August last year, China Cancer Society and other institutions issued the China Womens breast cancer screening guidelines which recommended that general risk women do breast X-ray screening every two years to achieve early diagnosis and treatment of breast cancer. At the same time, breast palpation is recommended as the initial method before breast imaging screening for women who have received breast cancer screening for the first time.

Source: editor in charge of interface news: Liao ziyao, nbjs10040