Nav: Home

Deep learning can distinguish recalled-benign mammograms from malignant and negative images

October 11, 2018

Bottom Line: An artificial intelligence (AI) approach based on deep learning convolutional neural network (CNN) could identify nuanced mammographic imaging features specific for recalled but benign (false-positive) mammograms and distinguish such mammograms from those identified as malignant or negative.

Journal in Which the Study was Published: Clinical Cancer Research, a journal of the American Association for Cancer Research.

Author: Shandong Wu, PhD, assistant professor of radiology, biomedical informatics, bioengineering, intelligent systems, and clinical and translational science, and director of the Intelligent Computing for Clinical Imaging lab in the Department of Radiology at the University of Pittsburgh, Pennsylvania

Background: "In order to catch breast cancer early and help reduce mortality, mammography is an important screening exam; however, it currently suffers from a high false recall rate," said Wu. "These false recalls result in undue psychological stress for patients and a substantial increase in clinical workload and medical costs. Therefore, research on possible means to reduce false recalls in screening mammography is an important topic to investigate."

How the Study Was Conducted: Wu and colleagues studied whether a technique in artificial intelligence called deep learning could be applied to analyze a large set of mammograms in order to distinguish images from women with a malignant diagnosis, images from women who were recalled and were later determined to have benign lesions (false recalls), and images from women determined to be breast cancer-free at the time of screening.

"The assumption is that there may be some nuanced imaging features associated with some mammogram images that could lead to a false/unnecessary recall when the images are interpreted by human radiologists, and our goal is to utilize a deep learning CNN-based method to build a computer toolkit to identify those potential mammogram images," Wu said.

The researchers used a total of 14,860 images of 3,715 patients from two independent mammography datasets, Full-Field Digital Mammography Dataset (FFDM - 1,303 patients) and Digital Dataset of Screening Mammography (DDSM - 2,412 patients). They built CNN models and utilized enhanced model training approaches to investigate six classification scenarios that would help distinguish images of benign, malignant, and recalled-benign mammograms.

Results: When the datasets from FFDM and DDSM were combined, the area under the curve (AUC) to distinguish benign, malignant, and recalled-benign images ranged from 0.76 to 0.91. The higher the AUC, the better the performance, with a maximum of 1, Wu explained. "AUC is a metric that summarizes the comparison of true positives against false positives, so it gives an indication not only of accuracy (how many were correctly identified), but also how many were falsely identified," he said.

Author's Comments: Wu said, "We showed that there are imaging features unique to recalled-benign images that deep learning can identify and potentially help radiologists in making better decisions on whether a patient should be recalled or is more likely a false recall."

"Based on the consistent ability of our algorithm to discriminate all categories of mammography images, our findings indicate that there are indeed some distinguishing features/characteristics unique to images that are unnecessarily recalled," Wu noted. "Our AI models can augment radiologists in reading these images and ultimately benefit patients by helping reduce unnecessary recalls."

Study Limitations: As limitations of the study, Wu noted that additional independent datasets could help further evaluate the accuracy and robustness of the algorithms, and utilizing alternative deep learning models, architectures, and model training strategies can help improve performance.
-end-
Funding & Disclosures: This study was supported by the National Institutes of Health, a Radiological Society of North America Research Scholar Grant, a University of Pittsburgh Cancer Institute Precision Medicine Pilot Award from The Pittsburgh Foundation, a University of Pittsburgh Physicians Academic Foundation Award, and donation of the Titan X Pascal GPU by Nvidia Corporation. Wu declares no conflicts of interest.

American Association for Cancer Research

Related Breast Cancer Articles:

Does MRI plus mammography improve detection of new breast cancer after breast conservation therapy?
A new article published by JAMA Oncology compares outcomes for combined mammography and MRI or ultrasonography screenings for new breast cancers in women who have previously undergone breast conservation surgery and radiotherapy for breast cancer initially diagnosed at 50 or younger.
Blood test offers improved breast cancer detection tool to reduce use of breast biopsy
A Clinical Breast Cancer study demonstrates Videssa Breast can inform better next steps after abnormal mammogram results and potentially reduce biopsies up to 67 percent.
Surgery to remove unaffected breast in early breast cancer increases
The proportion of women in the United States undergoing surgery for early-stage breast cancer who have preventive mastectomy to remove the unaffected breast increased significantly in recent years, particularly among younger women, and varied substantially across states.
Breast cancer patients with dense breast tissue more likely to develop contralateral disease
Breast cancer patients with dense breast tissue have almost a two-fold increased risk of developing disease in the contralateral breast, according to new research from The University of Texas MD Anderson Cancer.
Some early breast cancer patients benefit more from breast conservation than from mastectomy
Breast conserving therapy (BCT) is better than mastectomy for patients with some types of early breast cancer, according to results from the largest study to date, presented at ECC2017.
One-third of breast cancer patients not getting appropriate breast imaging follow-up exam
An annual mammogram is recommended after treatment for breast cancer, but nearly one-third of women diagnosed with breast cancer aren't receiving this follow-up exam, according to new findings presented at the 2016 Annual Clinical Congress of the American College of Surgeons.
Low breast density worsens prognosis in breast cancer
Even though dense breast tissue is a risk factor for breast cancer, very low mammographic breast density is associated with a worse prognosis in breast cancer patients.
Is breast conserving therapy or mastectomy better for early breast cancer?
Young women with early breast cancer face a difficult choice about whether to opt for a mastectomy or breast conserving therapy (BCT).
Breast density and outcomes of supplemental breast cancer screening
In a study appearing in the April 26 issue of JAMA, Elizabeth A.
Full dose radiotherapy to whole breast may not be needed in early breast cancer
Five years after breast-conserving surgery, radiotherapy focused around the tumor bed is as good at preventing recurrence as irradiating the whole breast, with fewer side effects, researchers from the UK have found in the large IMPORT LOW trial.

Related Breast Cancer Reading:

Dear Friend: Letters of Encouragement, Humor, and Love for Women with Breast Cancer
by Gina L Mulligan (Author)

Chicken Soup for the Soul: Hope & Healing for Your Breast Cancer Journey: Surviving and Thriving During and After Your Diagnosis and Treatment
by Dr. Julie Silver (Author)

Dr. Susan Love's Breast Book (A Merloyd Lawrence Book)
by Susan M. Love (Author), Karen Lindsey (Contributor), Elizabeth Love (Contributor)

Heal Breast Cancer Naturally: 7 Essential Steps to Beating Breast Cancer
by Dr. Veronique Desaulniers (Author)

The Breast Cancer Survival Manual, Sixth Edition: A Step-by-Step Guide for Women with Newly Diagnosed Breast Cancer
by John Link M.D. (Author), James Waisman M.D. (Author), Nancy Link R.N. (Author)

The Whole-Food Guide for Breast Cancer Survivors: A Nutritional Approach to Preventing Recurrence (The New Harbinger Whole-Body Healing Series)
by Edward Bauman MEd PhD (Author), Helayne Waldman MS EdD (Author), Donald I. Abrams MD (Foreword)

Breast Cancer Smoothies: 100 Delicious, Research-Based Recipes for Prevention and Recovery
by Daniella Chace (Author)

Just Get Me Through This! - Revised and Updated: A Practical Guide to Coping with Breast Cancer
by Deborah A. Cohen (Author), Robert M. Gelfand M.D. (Author)

Chicken Soup for the Breast Cancer Survivor's Soul: Stories to Inspire, Support and Heal (Chicken Soup for the Soul)
by Jack Canfield (Author), Mark Victor Hansen (Author), Mary Olsen Kelly (Author)

Best Science Podcasts 2018

We have hand picked the best science podcasts for 2018. Sit back and enjoy new science podcasts updated daily from your favorite science news services and scientists.
Now Playing: TED Radio Hour

Circular
We're told if the economy is growing, and if we keep producing, that's a good thing. But at what cost? This hour, TED speakers explore circular systems that regenerate and re-use what we already have. Guests include economist Kate Raworth, environmental activist Tristram Stuart, landscape architect Kate Orff, entrepreneur David Katz, and graphic designer Jessi Arrington.
Now Playing: Science for the People

#503 Postpartum Blues (Rebroadcast)
When a woman gives birth, it seems like everyone wants to know how the baby is doing. What does it weigh? Is it breathing right? Did it cry? But it turns out that, in the United States, we're not doing to great at asking how the mom, who just pushed something the size of a pot roast out of something the size of a Cheerio, is doing. This week we talk to anthropologist Kate Clancy about her postpartum experience and how it is becoming distressingly common, and we speak with Julie Wiebe about prolapse, what it is and how it's...