Breast cancer is one of the most common cancers in women, with more than 2 million new diagnoses last year alone. As we all know, early diagnosis and treatment is vital for cancer patients, so the accuracy of early diagnosis plays an important role.
In Britain, women over 50 are recommended to get a mammogram every 3 years, and the results of the mammogram are analyzed by 2 experts independently. However, interpreting the scans can go wrong, so a small percentage of all mammograms either return a false positive (misdiagnosing a healthy patient as having cancer) or false negative (missing the disease as it spreads) result.
This is when researchers at Google Health come in. They have trained an artificial intelligence model to detect cancer in breast scans from thousands of women in Britain and the US, and it is proved to work better than humans now.
Through an experiment, the AI model could predict breast cancer from the scans with a similar accuracy level to expert radiographers, and showed a reduction in the proportion of cases where cancer was incorrectly identified.
The team has conducted experiments comparing the computer’s decision with that of the first human scan reader (as I’ve said before, normally the scan interpreting requires two doctors.) If the diagnoses of AI and the first reader agree, the case is marked resolved. Only when the results are different is a second reader involved.
The results showed that using AI to verify the first human expert reviewer’s diagnosis could save up to 88% of the workload for the second clinician.
The team said further research was needed but they hoped that the technology could one day act as a “second opinion” for cancer diagnoses.