Currently radiologists must scan a tumour and then painstakingly draw the outline of the cancer on dozens of sections by hand to create a 3D map before treatment, a process which can take up to four hours.
They also must outline nearby important organs to make sure they are protected from the blast of radiation.
But Microsoft engineers have developed a programme which can delineate a tumour within minutes, meaning treatment can happen immediately.
The programme can also show doctors how effective each treatment has been, so the dose can be altered depending on how much the tumour has been shrunk.
“Eyeballing works very well for diagnosing,” said Antonio Criminisi, a machine learning and computer vision expert who heads radiomics research in Microsoft’s Cambridge, UK, lab.
“Expert radiologists can look at an image – say a scan of someone’s brain – and be able to say in two seconds, ‘Yes, there’s a tumor. No, there isn’t a tumor. But delineating a tumour by hand is not very accurate.”
The system could eventually evaluate 3D scans pixel by pixel to tell the radiologist exactly how much the tumor has grown, shrunk or changed shape since the last scan.
It also could provide information about things like tissue density, to give the radiologist a better sense of whether something is more likely a cyst or a tumor. And it could provide more fine-grained analysis of the health of cells surrounding a tumor.
“Doing all of that by eye is pretty much impossible,” added Dr Criminisi.
The images could also be 3D printed so that surgeons could practice a tricky operation, such as removing a hard-to -reach brain tumour, before surgery.