Machine learning identifies breast lesions likely to become cancer

October 17, 2017, Radiological Society of North America
Flow diagram shows patient selection. Credit: Radiological Society of North America

A machine learning tool can help identify which high-risk breast lesions are likely to become cancerous, according to a new study appearing online in the journal Radiology. Researchers said the technology has the potential to reduce unnecessary surgeries.

High-risk breast lesions are biopsy-diagnosed lesions that carry an increased risk of developing into . Because of that risk, surgical removal is often the preferred treatment option. However, many high-risk lesions do not pose an immediate threat to the patient's life and can be safely monitored with follow-up imaging, sparing the costs and complications associated with surgery.

"There are different types of high-risk lesions," said study author and radiologist Manisha Bahl, M.D., M.P.H., from Massachusetts General Hospital (MGH) and Harvard Medical School, both in Boston. "Most institutions recommend surgical excision for high-risk lesions such as , for which the risk of upgrade to cancer is about 20 percent. For other types of high-risk lesions, the risk of upgrade varies quite a bit in the literature, and patient management, including the decision about whether to remove or survey the lesion, varies across practices."

Dr. Bahl and colleagues at MGH studied the use of a tool to identify high-risk lesions that are at low risk for upgrade to cancer. The study resulted from a close collaboration between researchers at the Massachusetts Institute of Technology's (MIT) Computer Science and Artificial Intelligence Laboratory in Cambridge, Mass., and breast imaging experts at MGH.

Scatterplot shows score output of machine learning model plotted against a random number in the independent test set. Red circles represent high-risk breast lesions (HRL) upgraded to malignancy at surgery, and blue crosses represent HRLs not upgraded to malignancy at surgery. Vertical dotted line indicates 5 percent threshold, below which only one HRL was upgraded to malignancy at surgery. Credit: Radiological Society of North America
"Because diagnostic tools are inexact, there is an understandable tendency for doctors to over-screen for breast cancer," said coauthor Regina Barzilay, Ph.D., the Delta Electronics Professor of Electrical Engineering and Computer Science at MIT. "When there's this much uncertainty in data, machine learning is exactly the tool that we need to improve detection and prevent overtreatment."

Machine learning is a type of in which a model automatically learns and improves based on previous experiences. The model developed by researchers analyzed traditional risk factors such as patient age and lesion histology, along with several unique features, including words that appear in the text from the biopsy pathology report. The researchers trained the model on a group of patients with biopsy-proven high-risk lesions who had surgery or at least two-year imaging follow-up. Of the 1,006 high-risk lesions identified, 115, or 11 percent, were upgraded to cancer.

After training the machine learning model on two-thirds of the high-risk lesions, the researchers tested it on the remaining 335 lesions. The model correctly predicted 37 of the 38 , or 97 percent, that were upgraded to cancer. The researchers also found that use of the model would have helped avoid almost one-third of benign surgeries.

Graph shows accuracy achieved with machine learning model for independent test set as a function of model output score, both for patients with malignancy (red line) and for patients without malignancy (blue line), in independent test set. Vertical dotted line indicates 5 percent threshold. Credit: Radiologial Society of North America

The machine-learning identified the terms "severely" and "severely atypical" in the text of the pathology reports as associated with a greater risk of upgrade to cancer.

"Our study provides 'proof of concept' that machine learning can not only decrease unnecessary surgery by nearly one-third in this specific patient population, but also can support more targeted, personalized approaches to patient care," said the paper's senior author, Constance Lehman, M.D., Ph.D., professor at Harvard Medical School and Director of Breast Imaging at MGH.

"Our goal is to apply the tool in clinical settings to help make more informed decisions as to which patients will be surveilled and which will go on to surgery," Dr. Bahl added. "I believe we can capitalize on machine learning to inform clinical decision making and ultimately improve patient care."

Explore further: Study outlines recommendations for precancerous treatment of anal lesions

More information: "High-Risk Breast Lesions: A Machine Learning Model to Predict Pathologic Upgrade and Reduce Unnecessary Surgical Excision" Radiology, 2017.

Related Stories

Study outlines recommendations for precancerous treatment of anal lesions

September 27, 2017
Currently, no guidelines exist for screening and treatment of precancerous anal lesions or for anal cancer screening because enough is not known about the effectiveness of treating lesions detected through screening. A new ...

Surgical excision unnecessary in some patients with benign papillomas

May 3, 2012
Imaging surveillance is an acceptable alternative to surgical excision in patients with benign papilloma, diagnosed at breast core biopsy without cell abnormalities, a new study shows.

Some women with abnormal breast lesions may avoid surgery

July 30, 2013
Surgery is not always necessary for women with a type of breast tissue abnormality associated with a higher risk of cancer, according to a new study published online in the journal Radiology. Researchers said that periodic ...

Artificial intelligence helps with earlier detection of skin cancer

August 23, 2017
New technology being developed by researchers at the University of Waterloo and the Sunnybrook Research Institute is using artificial intelligence (AI) to help detect melanoma skin cancer earlier.

Recommended for you

Combining three treatment strategies may significantly improve melanoma treatment

December 12, 2018
A study by a team led by a Massachusetts General Hospital (MGH) investigator finds evidence that combining three advanced treatment strategies for malignant melanoma—molecular targeted therapy, immune checkpoint blockade ...

Lethal combination: Drug cocktail turns off the juice to cancer cells

December 12, 2018
A widely used diabetes medication combined with an antihypertensive drug specifically inhibits tumor growth—this was discovered by researchers from the University of Basel's Biozentrum two years ago. In a follow-up study, ...

Researchers use computer model to predict prostate cancer progression

December 12, 2018
An international team of cancer researchers from Denmark and Germany have used cancer patient data to develop a computer model that can predict the progression of prostate cancer. The model is currently being implemented ...

An integrated approach to finding new treatments for breast cancer

December 12, 2018
Unraveling the complexity of cancer biology can lead to the identification new molecules involved in breast cancer and prompt new avenues for drug development. And proteogenomics, an integrated, multipronged approach, seems ...

New insight into stem cell behaviour highlights therapeutic target for cancer treatment

December 12, 2018
Research led by the University of Plymouth and Technische Universität Dresden has identified a new therapeutic target for cancer treatment and tissue regeneration – a protein called Prominin-1.

Pushing closer to a new cancer-fighting strategy

December 11, 2018
A molecular pathway that's frequently mutated in many different forms of cancer becomes active when cells push parts of their membranes outward into bulging protrusions, Johns Hopkins researchers report in a new study. The ...

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.