x
Breaking News
More () »

Google AI could help increase detection of breast cancer

According to the American Cancer Society, mammograms generally fail to find 1 in 5 cases of breast cancer, but new technology is helping doctors detect missed d...

According to the American Cancer Society, mammograms generally fail to find 1 in 5 cases of breast cancer, but new technology is helping doctors detect missed diagnoses.

“The cancer was there, and we took a scan. But we essentially didn’t see it on the scan for whatever reason,” Dr. Mozziyar Etemadi said.

Northwestern University’s Dr. Mozziyar Etemadi refers to the mammogram’s risk of a false negative. According to the American Cancer Society, mammograms fail to find one in five breast cancers.

“Physicians can struggle to read some mammograms when the breast tissue is dense,” Dr. Laurie Margolies said.

Mount Sinai’s Chief of Breast Imaging, Dr. Laurie Margolies, explains how that fibrous tissue can obscure tumors from doctors, in her mind creating a need for new technology to reveal that hidden 20 percent of breast cancers.

“I think we’re excited about empowering radiologists with these tools, and hoping they become ubiquitous,” Scott McKinney said.

Google software engineer Scott McKinney helped design a program to use artificial intelligence to find tumors in mammograms.

A study published in the journal “Nature” found Google’s AI identified nine percent more cancer in screenings than its human counterparts, reducing false positives by nearly six percent.

“Using these tools allows clinicians to have significantly more free time to concentrate on what we all went to medical school for, which is to actually take care of patients,” Etemadi said.

“Bringing Google and other large players into the field can only be a positive,” Dr. Margolies said.

Dr. Margolies expects artificial intelligence to transform all of healthcare in the future, but especially radiology.

One in eight women are at risk of getting breast cancer in their lives. Doctors take 35 million mammograms every year, requiring massive computing power to combine and analyze all those images.

Researchers also found instances in which all six participating radiologists identified the cancer, while Google’s algorithm did not, suggesting programs like these might one day exist to supplement and not replace the human component.

“We’re excited by the synergy of putting the two together,” McKinney said.

Before You Leave, Check This Out