This Google program detects lung cancer faster than radiologists
AI and data analysis are the order of the day. The increasingly sophisticated algorithms and, created for all kinds of functions, have helped to improve processes and optimize work in many sectors, but they can still go to more. The last great discovery is that of a bot created by Google researchers that is capable of finding lung cancer with a higher percentage of success than a group of specialists.
So at least they have announced the Google AI researchers working on a project with the hospital manager Northwestern Medicine. His work was focused on creating an algorithm capable of detecting a lung cancer from screening tests performed by human doctors. After eight years of learning and development the first results have arrived.
According to an investigation published this week in the journal Nature Medicine, they analyzed the level of success by taking only one CT scan and saw that the model is right, on average, 5% more when it comes to finding a cancer that a group of six human experts and was 11% more likely to reduce false positives. Humans and AI obtained similar results when radiologists were able to see previous CT scans.
The algorithm was used to predict if a patient has lung cancer, generating a scoring system for analyze the risk that said tumor is malignant or not and identifying the location of the tissue in the lungs. The final idea is that this tool is available through the Google Cloud Healthcare API and that it can be used by specialists from around the world to improve the detection and analysis of these diseases.
"The artificial intelligence system uses 3D learning to analyze the complete anatomy in the CT scans of the thorax, as well as patches based on object detection techniques that identify regions with malignant lesions, "explain Google technical leader Shravya Shetty and product manager Daniel Tse.
The model was trained using more than 42,000 images of chest CT scan taken from almost 15,000 patients. The final idea is that it is much easier and faster to detect a disease that kills more than 2 million people a year.