Latest / "Isn't That Interesting?!": Breast Cancer Survivor Stories + Remarkable Research / Season 2 / Episode 9: Dr. Annalisa Occhipinti {Teesside University, United Kingdom}
Transcript
- 0:00Machine learning, or teaching computers how to think, takes
- 0:04more than just punching buttons. It takes data, a workforce, and,
- 0:08well, fundingwhenearlier.org heard about a researcher in the
- 0:12UK hoping to create an AI tool for an earlier breast cancer
- 0:17detection? It didn't hesitate to help.
- 0:20Here's Martha Kaley, founder and executive directorofearlier.org,
- 0:23with this week's edition of Isn't That Interesting?
- 0:27Well, welcome, Annalisa. I'm going to not try to
- 0:30pronounce the name because I know I'll get it wrong.
- 0:32It's so, but anyway, I'm saying this is Annalisa is here with a
- 0:37podcast with us and she's got so many different degrees.
- 0:42I, I, it would fill up a page, but she's going to introduce
- 0:45herself to you all and give you some update on the research that
- 0:51we helped fund for her and where that's going.
- 0:53And, and so I'll ask some questions along the way.
- 0:57I'm very curious. So go ahead and just start what
- 1:00you want to. Thank you.
- 1:02Thank you, Martha, for the introduction, for the
- 1:05invitation. It's a pleasure to be here
- 1:06today. Yeah, your pronunciation was
- 1:09great. My name is Annalisa and I'm a
- 1:12professor of computer science at Teesside University in the
- 1:15Northeast, where I work in the intersection between AI and
- 1:19biology. And actually my background is
- 1:23pure mathematics. I did my degree in Italy, in
- 1:27Sicily, and then I went to Cambridge for my PhD, where I
- 1:31did computational biology. And that's kind of where my
- 1:34journey started to apply at the moment.
- 1:38At the time was statistics, but then it turned it into AI in
- 1:43machine learning. So I started applying computer
- 1:45science, AI and mathematics into cancer biology and that's what
- 1:50took me here today to site where I mostly work in the application
- 1:56of AI for early detection of breast cancer.
- 1:59Well, I am so thrilled and I love the the blending of the
- 2:03science. I think that's really helpful.
- 2:06And and if you can kind of explain to us looking with AII
- 2:11think this is so I think it's opening doors that will help us
- 2:15explode with information. So if you could kind of describe
- 2:21how you see it working in conjunction with earlier
- 2:25detection of breast cancer. Yeah.
- 2:27There is actually a lot of work that is currently undergoing in
- 2:31the application of AI for early detection.
- 2:34And one of the most promising fields is the detection in in
- 2:40blood samples of what's called circulating tumor cells, which
- 2:45are cells they are leaving the main, the primary tumor site and
- 2:51then going through our blood vessels.
- 2:54And there is more and more research done at the moment to
- 2:58try to spot and identify these cells as soon as possible.
- 3:03And AI is helping in identify the utterance of the key
- 3:10informations that are available in those cells so that from the
- 3:15blood test we will be able to detect if any cancer cell is
- 3:19there and detect that as early as possible.
- 3:23Would you be detecting the protein markers that scientists
- 3:28are working on? Is is that what you'll be
- 3:30looking at? It's a pattern, right?
- 3:32It would be the pattern. Yeah, that that's exactly what
- 3:36we'll be looking at mostly will the the research is focusing on,
- 3:40it's called transcritomics, which goes through single cell
- 3:43MNA sequencing for example. And we will be looking at the
- 3:47genes and the expression of those genes.
- 3:50So we can think of genes as the instructions that are telling
- 3:53the cells what to do. And of course these instructions
- 3:56are different from cell to cell. So when we are able to identify
- 3:59a cell that's not behaving as we would have expected, we go one
- 4:04down and try to understand why that cell is not behaving as a
- 4:09normal cell and try to identify the mechanisms behind this
- 4:14behavior. So do you operate?
- 4:18Is it like do you work on blood samples?
- 4:21Is that what you try to do? Yeah, we the the future will
- 4:27definitely be in blood samples. But at the moment, since for AI
- 4:31we need a lot of data at the moment a lot of data comes from
- 4:34biopsy. So that's what biopsies.
- 4:37So that's what we are working on at the moment.
- 4:40So we got a panel of many patients.
- 4:44Most of our data comes from publicly available data sets,
- 4:48but also from our collaborators. And in those biopsies we tend to
- 4:53look at the way cells evolves over time and try to identify
- 4:58which cancer cell is at the earlier stage, which cancer
- 5:02cells are at later stages and trying to understand how those
- 5:07steps are happening. And again this happens again the
- 5:11transcriptomic level. Try to understand how those
- 5:14genes are guiding pain of those cells.
- 5:19So what about atypical cells do does that fall into the pattern
- 5:25that you're looking at or is it specifically had already moved
- 5:29into cancer cells? There are different algorithms
- 5:33that we can use to kind of infer where those cells are and then
- 5:39we would try to place them in the right order.
- 5:42So the idea is to use some key biomarkers that are telling us
- 5:47at what stage the cells are. And then by using these key
- 5:54biomarkers, which is like working with a puzzle, and then
- 5:57you look for the right pieces to place the right at the right
- 6:00place in your timeline. And then from that timeline, you
- 6:04start to place the sales in the right time.
- 6:08Can you tell me how the grant that we gave you, how that
- 6:12helped with your process? Yeah, the grant that you gave me
- 6:17was really helpful in developing the initial idea of applying AI
- 6:24for early breast cancer detection.
- 6:26So developing that initial idea into an actual model and that
- 6:32was pivotal even from my career. I got the grant a few years ago
- 6:36when I was probably a few months in my first independent position
- 6:41as a lecture. So it really, it really helped
- 6:44me moving my research forward. And since then, I've been able
- 6:48to expand on that model that I built through your brand, either
- 6:53applying to different cancer types, but also the knowledge
- 6:58that we developed through that brand are now applied in
- 7:01different projects. Also breast cancer immunotherapy
- 7:05since some of those key insights were related to how cells
- 7:10respond to different treatments. So looking at those mechanisms
- 7:15in parallel, through the grants, I got the opportunity to develop
- 7:18new connections with clinicians here in the UK, in the NHS, but
- 7:23also at the international level. So I've got a few collaborators
- 7:28in the US, Cornell University, Columbia, where we are working
- 7:32together and joining forces in the early detection of breast
- 7:36cancer. And finally, the branch is also
- 7:38helping me training new researchers into these fields.
- 7:43So we got a few PhD students that keep working fascinating
- 7:47fields of applying AI for early breast cancer detection.
- 7:52So it's been, it's been a great starting point for me.
- 7:55I bet you're just so excited to to see the progress.
- 8:00What do you think? What do you see for like 5 years
- 8:03down the road? I think in five years long roads
- 8:07we should be able to see the first early detection test being
- 8:12less invasive and more personalized.
- 8:15So it's a bit more on the working on blood samples how and
- 8:21even through imaging. So there is a lot of equipment,
- 8:26lab equipments being developed on the imaging side.
- 8:29So try to detect cancer at an earlier stage through different
- 8:35imaging techniques. So the the future of early
- 8:40detection I believe is going to be less invasive and more
- 8:44personalized Even so that not each single one of us will get
- 8:49detected for the same biomarkers, but it will depend
- 8:53on how history and how our biological fingerprints, let's
- 8:57say. So with the biomarkers, would
- 9:02you be able to assess the different types of breast cancer
- 9:07cells? I was speaking with Doctor Walt
- 9:10yesterday. So his blood test that he had,
- 9:14you know, he was looking for being able to, to define the
- 9:20subtypes. Does would would that comply
- 9:24with your work too? Yeah, exactly.
- 9:29Yeah. So the the work that we're doing
- 9:31for the subtypes is mostly focused on the metabolic
- 9:36modeling. So it's just another
- 9:39computational technique that tries to understand the
- 9:44metabolic changes of a cell. So when we talk about
- 9:47metabolism, we just think about how the cell uses the oxygen and
- 9:52the glucose in the environment in our body to survive.
- 9:56And actually cancer cells and healthy cells use these
- 10:00nutrients in totally different ways.
- 10:02And by using what we already know, we we model through
- 10:06metabolic modelling how this information can actually help
- 10:12cells grow. So how the the cancer cell uses
- 10:16these available nutrients differently from a normal cells.
- 10:21And by doing this, you can, you can also see how the differences
- 10:27are across different subtypes because with the basis that we
- 10:32have, we have cancer cells from different molecular subtypes.
- 10:36So we can actually see how different cells evolve and use
- 10:40the nutrients and across different subtypes and across
- 10:44different treatments. And as you can imagine,
- 10:46understanding mechanisms of how cell will survive or cell keeps
- 10:53growing and how developing targeted therapies to stop
- 10:58conscious cells proliferate or stop conscious cells to grow.
- 11:01So yeah, that's also something that we are currently working
- 11:04on. That's so interesting.
- 11:06So I know when they look liquid biopsies started or I became
- 11:12aware of them and the problem was you had it have such a large
- 11:17number circulating before they could be registered.
- 11:21But now with what you're doing, it will change that picture,
- 11:26right? Yeah, that's right.
- 11:28Even when I when I started working in this field, which was
- 11:31the beginning of my PSD in 2013, that was one of the main
- 11:34challenges. And when you were actually
- 11:37detecting the circulating tumor cells, it was kind of too late
- 11:40because already the tumor might have spread or was in the
- 11:44process of spreading to a secondary site.
- 11:46But nowadays with the better technology in terms of
- 11:49sequencing technology is much easier to detect those cells or
- 11:55even if I'm not circulating tumor, there are some proteins
- 12:00or some features that you can get from the blood samples that
- 12:04will tell you, will inform you that there is something that
- 12:08doesn't look quite right and that helps in progressing public
- 12:11screening and detecting the cancer at earlier stage.
- 12:17I think it's just so interesting to think about how it has all
- 12:22evolved in my lifetime. It is.
- 12:26I've been been doing this for 30 years and it's, it's just, you
- 12:31know, 30 years in a lifetime is a long time, but 30 years in
- 12:36science is just, it's just light years for us.
- 12:40So I, I really am excited about it.
- 12:42Do you have anything else that you would like to talk about?
- 12:46I think it's just as you mentioned, it's just a really
- 12:49exciting field to work on at the moment.
- 12:52And I'm really happy to be part of this new investigation, the
- 12:57application of AI in early detection, breast cancer, just
- 13:01because I think it's a really important field.
- 13:03And it's one of those areas where AI can really drive
- 13:08technology, but also drive the science forward.
- 13:11But I should also say that this wouldn't be impossible without
- 13:15everyone else involved. So it's not just AI, it's also
- 13:18the biologist, clinicians and everyone else around it.
- 13:21But it's just a, an effort, a common effort.
- 13:24And with work that you're doing at earlier.org, you're making
- 13:29this a common efforts even more possible and better.
- 13:32So thank you. Well.
- 13:34You're very welcome. Thank you.
- 13:36And I enjoy seeing all the different disciplines coming,
- 13:41being brought together so that it really gives such a strong
- 13:45force for for working on that. So thank you so much and I
- 13:50appreciate it and have and congratulations on being a
- 13:53professor. That's wonderful.
- 13:55Thank you. Thanks a lot for the
- 13:57opportunity. Thank you.
- 13:59OK, bye, bye. Episode of Isn't That
- 14:02Interesting was made possible thanks to volunteers and donors
- 14:06just like you. It was hosted by Martha Kaley,
- 14:09founderofearlier.org and edited by Liam Mcburney.
- 14:13I'm Valerie Amsterdam. Reminding you that for the last
- 14:1630 years, this is North Carolina based organization has helped
- 14:19fund groundbreaking research all over the world, from molecular
- 14:24imaging to a sensitive blood test for breast cancer.
- 14:27Plain and simple, earlier detection saves lives and you
- 14:30can help host a fundraising event.
- 14:33Make a donation for Volunteer your time.
- 14:35You can find us not only on the web but on Facebook and
- 14:38Instagram too. Let's change the picture of
- 14:41breast cancer together.