Google’s DeepMind division launched its Health initiative earlier this year in an effort to use machine learning for medical purposes. Now, that initiative has launched a project that would speed up planning for neck and cancer radiotherapy treatments. DeepMind Health has joined forces with the the UK’s National Health Service to analyze scans, all of which will be anonymized, from 700 former cancer patients at the University College London Hospital. See, it takes up to four hours for clinicians to painstakingly map out areas of the head and neck that need radiotherapy treatment. Those areas contain vital parts of the body, and clinicians need to make sure healthy cells remain untouched.
By analyzing samples from UCLH, DeepMind Health hopes to develop a technique to reduce the time it takes to map out areas that need treatment to an hour. It won’t make the process called “segmentation” automated. Clinicians will still be in charge, but it will make planning easier for medical professionals and will free up more of their time for patient care, education and research. DeepMind also intends to use the data it gleans from analyzing the samples to develop a radiotherapy segmentation algorithm that can be applied to other parts of the body.