Latest / DEEP INSIGHT / Doctors using AI mammogram readings to predict breast cancer risk

Transcript

Doctors using AI mammogram readings to predict breast cancer risk · DEEP INSIGHT

• 0:00
  Speaker 1: October is Breast Cancer Awareness month. It is the second
• 0:04
  Speaker 1: leading cause of cancer deaths in women. About one in
• 0:07
  Speaker 1: eight will be diagnosed with a disease in their lifetime,
• 0:10
  Speaker 1: and we're also seeing a rise in women under the
• 0:13
  Speaker 1: age of fifty being diagnosed. That's why early screening and
• 0:17
  Speaker 1: knowing if you're considered at risk can be life saving.
• 0:21
  Speaker 1: Mammogrants have saved countless lives, but they aren't perfect. Now
• 0:24
  Speaker 1: some doctors are actually using AI technology as a tool
• 0:29
  Speaker 1: to help them not only detect breast cancer, but also
• 0:32
  Speaker 1: predict a woman's risk factors. I want to welcome in
• 0:36
  Speaker 1: doctor Connie Lehman. She's a breast imaging specialist at mass
• 0:39
  Speaker 1: General Bingham and the founder of Clarity, the first FDA
• 0:44
  Speaker 1: authorized AI platform used to help predict a woman's five
• 0:48
  Speaker 1: year future risk of developing breast cancer. Doctor Lehman, thank
• 0:51
  Speaker 1: you so much for being with us. So how does
• 0:54
  Speaker 1: this technology work in detecting cancer before it's even formed.
• 1:00
  Speaker 2: I'm so excited about this. You know, there was there's
• 1:02
  Speaker 2: a radiologist way back in the nineteen seventies that thought
• 1:05
  Speaker 2: he noticed mammograms and women who end up being diagnosed
• 1:09
  Speaker 2: with breast cancer actually look different than mammograms and women
• 1:13
  Speaker 2: who don't develop breast cancer. He wondered, if we had
• 1:16
  Speaker 2: powerful enough computers, could we pull that information out. I
• 1:20
  Speaker 2: just wish you were alive today, because with AI we
• 1:22
  Speaker 2: can now do that. We can go far beyond breast density.
• 1:26
  Speaker 2: Breast density is important, we can hide cancers. We need
• 1:29
  Speaker 2: to be careful on how we talk to women about
• 1:31
  Speaker 2: their breast density, but it's actually a very weak predictor
• 1:35
  Speaker 2: of future breast cancer. And that's where AI comes in.
• 1:38
  Speaker 2: AI and the power of computer vision can extract many
• 1:42
  Speaker 2: more predictive elements than just the density and predict a
• 1:45
  Speaker 2: woman's future risk of breast cancer.
• 1:48
  Speaker 1: Forty percent of women have are told that they have
• 1:51
  Speaker 1: dense breast. Can you tell us a little bit what
• 1:53
  Speaker 1: that means to have a dense breast and how soon
• 1:56
  Speaker 1: we could see this technology in all medical facilities.
• 2:01
  Speaker 2: So all women's breast tissue has two basic components. One
• 2:05
  Speaker 2: is normal healthy breast fat and the other of these
• 2:09
  Speaker 2: fibrous and glandular elements, and those show up is white
• 2:12
  Speaker 2: on the mammogram as we can see here, whereas the
• 2:16
  Speaker 2: fat is more dark. It's like a gray color. And
• 2:19
  Speaker 2: cancers tend to be white, So people refer to it
• 2:22
  Speaker 2: as you're trying to find a white snowball and a
• 2:24
  Speaker 2: white snowstorm when you have a cancer and a very
• 2:27
  Speaker 2: dense breast So the masking the hiding of a cancer
• 2:31
  Speaker 2: can be really important. But also there's a little bit
• 2:33
  Speaker 2: of a bump up in risk if you have dense
• 2:36
  Speaker 2: breast tissue. Turns out that bump up is very, very small.
• 2:40
  Speaker 2: But when we use AI, we can really focus in
• 2:43
  Speaker 2: and target those women with dense breast issue who really
• 2:47
  Speaker 2: are at increase risk and reassure those that we find
• 2:50
  Speaker 2: are actually not at increase risk.
• 2:52
  Speaker 1: Now, there is some concern that AI may frankly be
• 2:55
  Speaker 1: too good at its job and they detect tumors that
• 2:58
  Speaker 1: are not life threatening. Patient may go through the painful
• 3:01
  Speaker 1: process of treating something that frankly didn't need to be
• 3:04
  Speaker 1: treated at all. What do you say to that concern.
• 3:07
  Speaker 2: Yeah, we've been struggling and working on and focused on
• 3:10
  Speaker 2: this idea of overdiagnosis finding cancers that don't matter, So
• 3:16
  Speaker 2: that's across the whole board. Obviously, our focus on is
• 3:20
  Speaker 2: to find cancers as early as possible when they can
• 3:22
  Speaker 2: be cured, but we do think when we have a
• 3:25
  Speaker 2: risk model like Clarity breast, we can actually reduce the
• 3:29
  Speaker 2: quote over diagnosis because we're going to find the right
• 3:33
  Speaker 2: kinds of patients to have the right kinds of test.
• 3:36
  Speaker 2: What I worry about is when we take expensive, new
• 3:40
  Speaker 2: invasive technologies and we suggest that all women need them,
• 3:44
  Speaker 2: or all women with dense breast tissue need them, when
• 3:48
  Speaker 2: in fact, we can be much more precise in how
• 3:50
  Speaker 2: we direct women in that decision making so that those
• 3:54
  Speaker 2: that are having more invasive, more aggressive screening and cancer
• 3:59
  Speaker 2: prevention interventions are the right women.
• 4:03
  Speaker 1: Right now, the guidelines say that you need to start
• 4:05
  Speaker 1: getting a mammogram or getting screened at the age of forty.
• 4:08
  Speaker 1: With this new technology and its ability to detect cancer
• 4:12
  Speaker 1: before it even starts, do you think that the guidelines
• 4:15
  Speaker 1: may change that women may start getting screened even earlier.
• 4:20
  Speaker 2: Absolutely. Until now, we really were very limited in how
• 4:25
  Speaker 2: we recommended women start screening mimography. We've been arguing for
• 4:29
  Speaker 2: a long time, is it fifty or forty. Most parts
• 4:31
  Speaker 2: of the world, women don't start screening until fifty. In
• 4:34
  Speaker 2: the US, we now have pretty good agreement that we
• 4:38
  Speaker 2: should start at forty. But what we know is the
• 4:42
  Speaker 2: fastest rising group of breast cancer diagnoses or in women
• 4:46
  Speaker 2: under forty. It's also the fastest rising group of metastatic
• 4:49
  Speaker 2: breast cancer that can't be cured. We know that shock
• 4:52
  Speaker 2: when we hear someone was diagnosed with breast cancer but
• 4:56
  Speaker 2: they were only thirty three. How is that possible? They
• 4:58
  Speaker 2: were only thirty six. So we've done a lot as
• 5:01
  Speaker 2: a healthcare community to identify women at high risks that
• 5:05
  Speaker 2: should start screening in the thirties. Do they have a
• 5:07
  Speaker 2: first degree relative with breast cancer diagnosed before the age
• 5:10
  Speaker 2: of fifty, Do they have a genetic mutation, But we're
• 5:14
  Speaker 2: missing the majority of women in their thirties being diagnosed
• 5:17
  Speaker 2: with breast cancer. We just completed a study showing that
• 5:20
  Speaker 2: we can identify ten percent of women one in ten
• 5:24
  Speaker 2: in their thirties that have a much higher risk of
• 5:27
  Speaker 2: future breast cancer than women were screening in their fifties
• 5:30
  Speaker 2: and sixties. So that's groundbreaking and that's where we really
• 5:34
  Speaker 2: see a future where we have a baseline at thirty,
• 5:37
  Speaker 2: we assess the woman's risk on that baseline mammogram at thirty,
• 5:41
  Speaker 2: we talked to her about her family history and her
• 5:44
  Speaker 2: other health history, and then we put her on a
• 5:46
  Speaker 2: plan to move safely through her thirties until she reaches
• 5:51
  Speaker 2: forty and starts average annual age forty, Average risk based.
• 5:56
  Speaker 1: Screening doctor Connie Leeman, Thank you so much.
• 6:00
  Speaker 2: Thank you