In fact, Google’s ambitions and projects in the field of health have been bundled and outsourced in the alphabet subsidiary Verily, but the researchers are still looking at this area. Now, research on artificial intelligence has found a new area of use that is likely to save a lot of human lives in the future: the AI is already able to recognize cancer better than a pathologist.
In recent years Google has taken a step further in research on artificial intelligence, and has repeatedly shown useful applications ranging from energy saving to the enlargement of photos.
In a white paper released today, the researchers show that a part of the artificial intelligence can also simplify a very important and difficult job in medicine: the diagnosis of cancer. Pathologists need to go through several images with a resolution of more than 10 gigapixels in particularly severe cases to detect this insidious disease. Since the recordings are magnified up to 40 times, this is a mammoth task not only costs a lot of time, but also requires full concentration.
Google has now run its own AI on these recordings, and was able to achieve a hit rate of 89 percent after only a short training phase and this within a very short time. The pathologist, on the other hand, had unlimited time and came only to a rate of 73 per cent. However, it must be said that Google’s Artificial Intelligence has often delivered false positives which is a catastrophe in the case of such a life-changing diagnosis.
The aim of the researchers, however, is not to make the profession of the pathologist superfluous, but to make it easier for him to work. The artificial intelligence is to recognize potential cancer cells and to present them to the physician, who then only needs to look closer at the individual marked areas. Since the diagnosis is based on a few other factors besides the pictures, it is important that not so much time is devoted to the evaluation of the material.
Interestingly, the technology for recognizing from the Self Driving Car project is actually designed to quickly perceive the environment and to recognize traffic, traffic signs and other moving things. This shows once again that research in such important areas always brings synergies to completely different areas. The early detection of cancer is another very good example of this, whereby the AI can be used in the future.