John McCarthy, a professor of computer science at Stanford who died in 2011, is widely known as the father of artificial intelligence. That accolade is partly due to the fact that, in the mid-1950s, McCarthy coined the phrase “Artificial Intelligence”.
McCarthy defined artificial intelligence (AI) as “the science and engineering of making intelligent machines”.
Today AI is recognized as the science which enables machines to accomplish tasks traditionally associated with human intelligence, including the abilities to learn and to solve problems.
AI has now reached the point where its potential to surpass human abilities is overwhelmingly obvious. The ability of AI to vastly outperform its human creators inspires both euphoria and dread. (Remember Skynet, the AI system that created Schwarzenegger’s character in the Terminator movies?)
AI screening mammograms for an accurate cancer diagnosis
One application of AI which definitely comes down on the euphoric side is the burgeoning use of AI in the detection of breast cancer.
Breast cancer is the type of cancer most often incurred by women worldwide. It’s the second leading cause of death among women.
Almost 270,000 new cases of breast cancer were reported in the US during 2019. The US death toll for that year is expected to reach nearly 42,000, according to the American Cancer Society. About one in eight American women will develop breast cancer during their lifetimes.
Globally, there are about 2 million new cases of breast cancer each year, and more than 500,000 deaths.
What is a false negative?
Early detection of breast cancer is of primary importance in effecting a cure. But the breast imaging tests currently in use have relatively high rates of error. About 20% of screenings fail to find existing breast cancer. This mistake is known as a false negative. The consequences of a false negative are both obvious and severe. Cancer that isn’t detected can’t be treated, and delay in detection can be fatal.
What is a false positive?
Over a 10-year span, 50% of women who get annual mammograms will receive a mistaken diagnosis that they have cancer. This sort of error is called a false positive. The downside of false positives is less severe than a false negative, but it’s still significant. False positives lead to additional imaging sessions, unnecessary biopsies, and inescapable concomitant anxiety.
Recent studies show that AI can reduce mistaken diagnoses of breast cancer
Last year, a team of scientists from New York University School of Medicine published a study demonstrating an AI system that equaled the accuracy of human radiologists in the detection of breast cancer.
Last June, researchers at Stanford reported that AI can help radiologists improve their interpretation of mammograms. The Stanford team used machine learning (a subset of artificial intelligence that teaches machines to teach themselves) to analyze 112,000 mammography cases collected from 13 radiologists at two teaching hospitals.
The radiologists’ performance was measured against the results obtained from the machine learning model. In 176 cases, the radiologists mistakenly stated that the patient was cancer-free (false negatives). In 12,476 cases, the radiologists mistakenly reported that cancer existed (false positives).
The AI machine learning algorithm reported one more false-negative than the humans did but reduced the false-positive reports by 3,612.
The Stanford team concluded that AI helps even expert radiologists significantly reduce false positives.
Now doctors can Google breast cancer
Google has recently launched several projects designed to show that AI is the diagnostic tool of medicine’s future.
The company has already created algorithms to help detect lung cancer and diagnose eye disease in diabetic people.
Now Google has shown that an AI system using deep learning (the step that follows machine learning in AI’s rapid evolution) is actually superior to human diagnosticians.
Google mammographic breast screening is not yet available for widespread use
But its unveiling seems to show that AI will inevitably assume a major role in the detection of cancer in breast tissue.
Google, in tandem with scientists from Northwestern University in Illinois and two British medical centers, let their AI deep learning system study mammograms from about 76,000 women in Britain and 15,000 in the United States.
Then the Google team tested the AI system on images from about 25,000 other British women, together with 3,000 mammograms from American women, and compared the AI’s performance with the radiologists who had originally read the mammograms.
Because the mammograms had been taken in the past, and the eventual outcomes of the women’s cases were therefore known, the accuracy rates of the diagnoses derived from those mammograms were also known.
With US mammograms, the system produced a 9.4% reduction in false negatives (a mammogram mistakenly read as non-cancerous) and a 5% reduction in false positives (an erroneous report that cancer does exist).
With British mammograms, the Google deep learning AI also surpassed the radiologists, reducing false negatives by 2.7% and false positives by 1.2%
More AI research is needed for breast cancer screening
All the scientists involved in each of these studies agree on one thing. There’s a lot more work to be done on screening programs before AI can be expected to accurately read each and every mammogram.
Tomosynthesis is a new technique for breast cancer detection
Digital breast tomosynthesis produces three-dimensional digital mammography using low-dose X-rays and computer reconstruction. This new medical imaging process has already yielded yet another level of diagnostic accuracy.
When the new deep learning AI systems are trained on a sufficient database of tomosynthetic images, it may be that AI will become the frontline warrior in the diagnostic battle against breast cancer. Until that time comes, AI can be expected to be at least an extremely valuable adjunct to the skill of human radiologists. Let me know what you think! Thaïs
About Dr. Thaïs Aliabadi
As one of the nation’s leading OB-GYNs, Dr. Thaïs Aliabadi offers the very best in gynecological and obstetric care. Together with her warm professional team, Dr. Aliabadi supports women through all phases of life. She fosters a special one-on-one relationship between patient and doctor.
We invite you to establish care with Dr. Aliabadi. Please click here to make an appointment or call us at (844) 863-6700.
We take our patients’ safety very seriously. Our facility’s Covid-19 patient safety procedures exceed all CDC and World Health Organization recommendations. Masks are required in our office at all times during the coronavirus pandemic.
The practice of Dr. Thais Aliabadi and the Outpatient Hysterectomy Center is conveniently located to patients throughout Southern California and the Los Angeles area. We are near Beverly Hills, West Hollywood, Santa Monica, West Los Angeles, Culver City, Hollywood, Venice, Marina del Rey, Malibu, Manhattan Beach, and Downtown Los Angeles, to name a few.