Google has won a contract with the Defense Innovation Division (DIU), part of the U.S. Department of Defense, to use the latest technological advances to improve the accuracy of cancer diagnoses. The current rate of misdiagnosis in the United States is about 5%, and half of these cases cause more problems.
Photo from pixnio.
Google plans to use the open source platform TensorFlow to train an artificial intelligence to detect cancer cells in photos taken under a microscope. They have begun using Google’s Cloud Healthcare API to develop neural networks that identify and segment existing data sets. Once AI is fully trained, they will design their own microscope and integrate AR (augmented reality) overlays to show doctors information about the likelihood of cell cancer.
Mike Daniels, vice president of global public sector at Google Cloud, said: “Speed and accuracy are critical to effectively treating cancer. We are working with DIU to help front-line healthcare practitioners understand the capabilities that can improve the lives of our military and their families by providing our machine learning and artificial intelligence technologies. “
Google hopes their approach will reduce the “huge amount of data” doctors face and make diagnosis faster, cheaper and more accurate. But it will take some time for artificial intelligence to be widely used in medical practice.
The first microscopes will be shipped to selected Defense Health Agency treatment facilities for research purposes only. The technology will then be rolled out to a wider range of U.S. military health systems and Veterans Affairs hospitals for practical trials.