Stanford University’s Computer Program Battle Against Breast Cancer

by on November 11, 2011

in bruce's thoughts

Stanford University

Daphne Koller, a computer science professor at Stanford University, is the senior author on a study that looks at how a computer program can examine the pathology of breast cancer images and predict at a much greater success rate, a patients’ survival chances from breast cancer.

Right now, the method which has been used for over 80 years, is still the main stream focus of pathology for doctors reviewing tumors under a microscope and ranking or scoring what they see from an established scale.

The scores determine the type and diversity of the cancer they’re battling.

This is the information that is used to calculate the patient’s outlook and course of their treatment.

The Stanford study suggests that the computer models they’ve been studying are finding additional factors that could indicate or influence survival rates.

The computer modeling is called Computational Pathologist, or C-Path.

Once the computer was trained in how to further analyze biopsied breast tissue, it could take the analysis a step farther and determine what cancer features matter the most or least in predicting the patient’s survival.

They gave the program six-thousand different cellular factors to review and one of the aspects that came out of the study was that there were characteristics of surrounding cells that can impact the cancer’s environment.  These factors are additional measures in predicting survival.

Breast Cancer Awareness Month pink ribbon symbolOf those 6,000 factors, there were at least eleven that ultimately indicated the most robust affiliations with a patient surviving the cancer.

Studies conducted on known test cases showed a significant increase over human-based reviews.

As it stands right now, C-Path is not going to replace pathologists, but down the road, will be a complimentary tool.  No commercial development from C-Path has begun yet, but this is research, as they said, is “landmark” work, even if the study right now has a limited control system.  But it’s an objective, quantitative system that can further back up or inform the human pathologists in their work.

And that can never be a bad thing.

The study is published Nov. 9 in Science Translational Medicine.

usatoday:  Computer-pathologist-could-assess-breast-cancer-survival  pathology_report

[Palo Alto 11/10/11 Daily Post, pg 12]

{ 0 comments… add one now }

Leave a Comment

Previous post:

Next post: