Stanford University researchers have trained computers to analyze microscopic images of breast cancer with greater accuracy than evaluations conducted by humans, the university said.
Their model, called Computational Pathologist, or C-Path, is a machine-learning-based method for analyzing images of cancerous tissues and predicting patient survival.
Traditionally, pathologists have examined tumors visually and scored them according to a scale first developed 80 years ago, helping them assess the type and severity of the cancer, course of treatment and patient prognosis.
To train C-Path, researchers used existing tissue samples taken from patients whose prognoses were known. The computers analyzed the images, measuring tumor structures and using the information to predict survival. By comparing results against the known data, the computer adapted their models to improve their predictions and gradually figured out which features of the cancers matter most and least relative to survival.
"In essence, the computer learns," said Daphne Koller, professor of computer science and author of a paper published Nov. 9 in "Science Translational Medicine."
"Pathologists have been trained to look at and evaluate specific cellular structures of known clinical importance, which get incorporated into the grade," said physician Andrew Beck, a doctoral candidate in bioinformatics and the paper's first author.
"However, tumors contain innumerable additional features, whose clinical significance has not previously been evaluated," Beck said.
"The computer strips away that bias and looks at thousands of factors to determine which matter most in predicting survival," Koller said.
C-Path assesses 6,642 cellular factors. Once trained using one group of patients, C-Path was asked to evaluate tissues of cancer patients it had not checked before and the result was compared against known data. Ultimately, C-Path yielded results that were a statistically significant improvement over human-based evaluation.
The computers identified structural features in cancers that matter as much or more than those that pathologists have focused on traditionally. In fact, they discovered that the characteristics of the cancer cells and the surrounding cells, known as the stroma, were both important in predicting patient survival.
The Stanford findings add weight to what many scientists have been contending for some time: that cancer is an "ecosystem," and that clinically significant information can be obtained by analysis of the complete tumor microenvironment.
"Through machine learning, we are coming to think of cancer more holistically, as a complex system rather than as a bunch of bad cells in a tumor," said Matt van de Rijn, a professor of pathology and co-author of the study.
"The computers are pointing us to what is significant, not the other way around."
Van de Rijn does not see computers replacing pathologists. "We're looking at a future where computers and humans collaborate to improve results for patients across the world," he said.
The researchers said the results of their work will be felt broadly and individually. Having computers that can evaluate cancers could bring world-class pathology to underserved areas where trained professionals have traditionally been scarce, improving the prognosis and treatment of breast cancer for millions in developing areas of the world.
Machine-learning also could reduce the variability in results. C-Path could improve the accuracy of prognoses for all breast cancer patients and also improve the screening of pre-cancerous cells that could help many women avoid cancer altogether. It might even be applied to predict the effectiveness of various forms of treatment and drug therapies.
"If we can teach computers to look at a tumor tissue sample and predict survival, why not train them to predict from the same sample which courses of treatment or drugs a given patient might respond to best? Or even to look at samples of non-malignant cells to predict whether these benign tissues will turn cancerous," said Koller. "This is personalized medicine."
Stanford Instructor of Pathology Ankur Sangoi; Research Associate Robert Marinelli and Associate Professor of Pathology Robert West also contributed to the study.