Computer accurately identifies and delineates breast cancers on digital tissue slides

May 10, 2017, Case Western Reserve University
A tumor boundary delineated by a pathologist. Credit: Anant Madabhushhi

A deep-learning computer network developed through research led by Case Western Reserve University was 100 percent accurate in determining whether invasive forms of breast cancer were present in whole biopsy slides.

Looking closer, the correctly made the same determination in each individual pixel of the 97 percent of the time, rendering near-exact delineations of the tumors.

Compared to the analyses of four pathologists, the machine was more consistent and accurate, in many cases improving on their delineations.

In a field where time and accuracy can be critical to a patient's long-term prognosis, the study is a step toward automating part of biopsy analysis and improving the efficiency of the process, the researchers say.

Currently, cancer is present in one in 10 biopsies ordered by physicians, but all must be analyzed by pathologists to identify the extent and volume of the disease, determine if it has spread and whether the patient has an aggressive or indolent cancer and needs chemotherapy or a less drastic treatment.

Last month, the U.S. Food and Drug Administration approved software that allows pathologists to review biopsy slides digitally to make diagnosis, rather than viewing the tissue under a microscope.

"If the network can tell which patients have cancer and which do not, this technology can serve as triage for the pathologist, freeing their time to concentrate on the cancer patients," said Anant Madabushi, F. Alex Nason professor II of biomedical engineering at Case Western Reserve and co-author of the study detailing the network approach, published in Scientific Reports.

A tumor delineated by deep-learning network. Credit: Anant Madabhushi

The study

To train the deep-learning network, the researchers downloaded 400 biopsy images from multiple hospitals. Each slide was approximately 50,000 x 50,000 pixels. The computer navigated through or rectified the inconsistencies of different scanners, staining processes and protocols used by each site, to identify features in cancer versus the rest of the tissue.

The researchers then presented the network with 200 images from The Cancer Genome Atlas and University Hospitals Cleveland Medical Center. The network scored 100 percent on determining the presence or absence of cancer on whole slides and nearly as high per pixel.

"The network was really good at identifying the cancers, but it will take time to get up to 20 years of practice and training of a pathologist to identify complex cases and mimics, such as adenosis," said Madabhushi, who also directs the Center of Computational Imaging and Personalized Diagnostics at Case Western Reserve.

Network training took about two weeks, and identifying the presence and exact location of cancer in the 200 slides took about 20 to 25 minutes each.

That was done two years ago. Madabhushi suspects training now—with new computer architecture—would take less than a day, and identification and delineation could be done in less than a minute per slide.

"To put this in perspective," Madabhushi said, "the machine could do the analysis during 'off hours,' possibly running the analysis during the night and providing the results ready for review by the pathologist when she/he were to come into the office in the morning."

Explore further: Computerized tissue image analysis reveals underlying genomics of ER+ breast cancer

More information: Angel Cruz-Roa et al, Accurate and reproducible invasive breast cancer detection in whole-slide images: A Deep Learning approach for quantifying tumor extent, Scientific Reports (2017). DOI: 10.1038/srep46450

Related Stories

Computerized tissue image analysis reveals underlying genomics of ER+ breast cancer

September 7, 2016
The number of tubules in tumors may predict which women with estrogen receptor positive (ER+) breast cancer will benefit from hormone therapy alone and which require chemotherapy, researchers at Case Western Reserve University ...

New computer program beats physicians at brain cancer diagnoses, could eliminate costly and risky brain biopsies

September 15, 2016
Computer programs have defeated humans in Jeopardy!, chess and Go. Now a program developed at Case Western Reserve University has outperformed physicians on a more serious matter.

Shape of prostate and compartments within may serve as cancer indicators

February 1, 2017
Preliminary computerized imaging reveals the shape of the prostate and a compartment within the gland—called the transitional zone—consistently differ in men with prostate cancer than those without the disease, according ...

Researchers developing quick, inexpensive test to assess ER+ breast cancers

July 11, 2016
Researchers from Case Western Reserve University are teaming with industry and other academics to develop a quick and inexpensive test to predict which women with ER+ breast cancer need chemotherapy and which need only the ...

Researchers building digital pathology tools to predict cancer outcomes

November 11, 2015
Case Western Reserve University researchers have been awarded two grants totaling $3.16 million from the National Institutes of Health (NIH) to create analytic software for managing, annotating, sharing and analyzing digital ...

Pathologists often disagree on breast biopsy results when diagnosing DCIS

March 21, 2016
A study applying B-Path (Breast Pathology) Study results to patient populations found that pathologists disagree with one another about 8 percent of the time when diagnosing a single breast biopsy slide. Discordance was more ...

Recommended for you

New drugs are improving survival times for patients with aggressive type of blood cancer, study finds

June 25, 2018
Survival times for a highly aggressive type of blood cancer have nearly doubled over the last decade due to the introduction of new targeted drugs, a Yorkshire study has shown.

Dying cancer cells make remaining glioblastoma cells more aggressive and therapy-resistant

June 21, 2018
A surprising form of cell-to-cell communication in glioblastoma promotes global changes in recipient cells, including aggressiveness, motility, and resistance to radiation or chemotherapy.

Existing treatment could be used for common 'untreatable' form of lung cancer

June 21, 2018
A cancer treatment already approved for use in certain types of cancer has been found to block cell growth in a common form of lung cancer for which there is currently no specific treatment available.

Novel therapy makes oxidative stress deadly to cancer

June 21, 2018
Oxidative stress can help tumors thrive, but one way novel cancer treatments work is by pushing levels to the point where it instead helps them die, scientists report.

Researchers uncover new target to stop cancer growth

June 21, 2018
Researchers at the University of Wisconsin-Madison have discovered that a protein called Munc13-4 helps cancer cells secrete large numbers of exosomes—tiny, membrane-bound packages containing proteins and RNAs that stimulate ...

Higher body fat linked to lower breast cancer risk in younger women

June 21, 2018
While obesity has been shown to increase breast cancer risk in postmenopausal women, a large-scale study co-led by a University of North Carolina Lineberger Comprehensive Cancer Center researcher found the opposite is true ...

0 comments

Please sign in to add a comment. Registration is free, and takes less than a minute. Read more

Click here to reset your password.
Sign in to get notified via email when new comments are made.