The new deep learning algorithm by Google, known as Google AI, is a new sensitive tool used to identify different types of breast cancer.
The sooner you can detect cancer, the better outcome you will have. However, diagnosing the different kinds of cancer in its early stages is often impossible. It takes years of education and training to detect cancerous tumors. And even with thorough training, symptoms may be harder to distinguish from other illnesses. For example, diagnosing some kinds of breast cancer is only 48% accurate.
To overcome this issue, Google has developed an efficient automated detection innovation to identify breast cancer. The program can scan thousands of cancer cell slides to recognize distinct patterns.
Although the technology is new and has a lot to prove in its early stages, Google believes their innovation is capable of matching or even exceeding a pathologist’s performance.
At this time, the algorithm is at 89% accuracy. Comparing it to the 73% of a pathologist, it is clearly more efficient and doesn’t require time constraints like that of a pathologist.
Google released a statement saying they aren’t the only ones seeing outstanding results. Other groups received 81% accuracy with the same data. Their model also generalized well, even to pictures from different hospitals with different scanners.
As with other deep learning algorithms, the Google AI isn’t close to perfect. It did great for the task, but still lacks certainty in detecting different abnormalities that it wasn’t trained to know. The algorithm must complement the workflow of the pathologists to ensure the best outcome.
Google researchers wrote that they picture an algorithm like theirs improving the consistency and efficiency of pathologists. They can also allow pathologists to accurately and easily measure the size of the tumor, a major factor when diagnosing patients with cancer.