Narrowed since the beginning of the 20th century, pathologists have had an eye to the microscope, searching for some of the features in the sample in a biopsy of the tumor that allows them to classify how aggressive the cancer is.
Information that helps doctors decide how to treat patients.
Today, armed with sophisticated software, powerful computers get quite skilled at pattern recognition. Identify the face, for example. Stanford researchers think the computer might be able to learn to evaluate cancer biopsy, as well.
To do this, Daphne Koller and his colleagues began with biopsy slides that are used to train pathologists. Slide scanned into the computer, which is measured not just some of the features of human pathologists can review, but thousands of characteristics of each image.
"And we are connected to the machine learning algorithms that look at data's survival," he said, "and trying to figure out which of these features is a good feature for live, which features the bad, and that is not relevant at all.
In fact, Koller said, the system identified previously known features on biopsy slides that helps predict how aggressive the cancer will.
"It turns out that some of the most significant features of the tumor that pathologists do not see at all now."
After the training phase, "C" streets "(computing Pathologist) evaluated both the pathology slide sets, coming from various hospitals and the medical and demographic is different from the training group. Actually it's better than humans, although pathologists Koller stressed that C-Path is not designed to replace the doctor saw through the microscope.
He said the computerized systems could have the greatest impact in resource-poor settings – in developing countries, for example — where skilled pathologists in short supply.
"Our technology can be easily implemented even over the Web, where local doctor extracts the sample and place it on the slide and scanned into the computer and send it over the net, and out comes the prediction of survival and finally the other aspects of samples that can help guide treatment."
Koller and her colleagues describe the computerized systems in Translational Medicine pathologists. They are currently working to extend the computerized system of other cancer pathology. It is still a few years of use in the treatment of patients.
People with life-threatening disease, preserve and share their stories
From archaeology to Zoology, from atoms to galaxies, Our World gives a mixture of state-of-the-art news and features