Google has shown more evidence that AI can become a valuable tool in spotting cancer. Google’s researchers have developed a deep learning tool that can detect advanced stage (metastatic) of breast cancer more accurately than pathologists, when studying the slides.
The team of researchers trained the deep learning algorithm, named as Lymph Node Assistant (LYNA), to identify the features of tumors by means of two groups of pathological slides. Thus, making it capable of spotting metastasis in many different conditions.
The work resulted in an Artificial Intelligence (AI) system that can distinguish between cancer and non-cancer slides with up to 99 percent accuracy, even when searching for very small metastases that might skip human’s notice.
LYNA was even more accurate when working as an ally – pathologists discovered that deep learning tech has made it easier for them to perform simulated diagnosis. It not only decreased the rate of skipped micro-metastases by a “factor of two”, it also decreased the inspection time to a single minute by dividing it in half.
Although this is a great step forward, this tool only provides limited help, as it can only identify the late-stage of breast cancer which doesn’t have any known cure. Furthermore, it still has to be applied to real-life medical conditions.
Nevertheless, scientists have noticed that the metastasis stage is the same for different forms of cancer. It wouldn’t require much effort to also configure the AI system to identify other tumors. Though, when it would be ready for real use, it could result in both very dependable diagnoses and allow doctors to use more of their time to take care of their patients.
