Dr. Veronica Rotemberg and her colleagues are working to address bias that exists within imaging data sets, which are used in training AI tools to recognize potential cancers.
Veronica Rotemberg, MD, PhD, Director of Dermatology Informatics Program at Memorial Sloan Kettering Cancer Center (MSK), and leader of her own lab at Memorial Hospital, is passionate about her patients, technology, and the future. She has been a member of the MSK faculty for five and a half years and is committed to advancing tools and technology to help provide optimal care for patients.
“My research focuses on AI [Artificial Intelligence] and imaging for skin cancer detection,” she explains. “We are intent on developing tools that catch skin cancers early, when they can still be cured with surgery and before they have spread beyond the skin.”
Teaching Technology
Skin cancer is the most common cancer — and one of the most treatable when discovered early. AI offers doctors an unprecedented ability to find possible skin cancers. Without AI, dermatologists often biopsy many more benign lesions than malignancies.
“It’s a risk we accept because we really want to find all the skin cancers that we possibly can,” Dr. Rotemberg says. One of the goals of her research is to make AI tools smarter and more aware of biological nuances. Dr. Rotemberg and her team are “training” AI tools to identify suspicious lesions with such precision that they will help eventually to reduce the number of biopsies of benign growths.
Training AI tools to serve as technological partners in identifying skin cancer involves extensive testing in the real-world context of patient care.
“Doctors don’t just use pictures to come up with a diagnosis,” Dr. Rotemberg says. “They talk to the patient, they feel the lesion, they ask questions, etc. Testing the tools in context is one of the big research projects of our group.”
Dr. Rotemberg and her colleagues are also working to address bias that exists within imaging data sets, which are used in training AI tools to recognize potential cancers.
“We know that imaging data sets are not very diverse or even labeled for things that might contribute to bias,” Dr. Rotemberg says. “Most public images don’t contain any labels for ethnicity, race, or skin tone, for example, which could affect algorithm performance. It’s an issue that we’re exploring.”
Dr. Rotemberg sees AI as a powerful ally to help physicians provide better care to everyone who needs it. She’s aware of the highly publicized fears that AI could be used in harmful ways. But Dr. Rotemberg’s primary concern are the inequities suffered by many people right now, rather than the possible threats posed by technology in the future.
“When we worry about the existential threat of AI, it makes it easy for us to ignore the harms that are currently happening to marginalized populations,” Dr. Rotemberg says. “I don’t stay up at night worrying that AI is going to cause the apocalypse. What keeps me up is thinking about racial bias in medical AI algorithms that could determine who gets certain treatments or who gets access to a certain specialist. I think the possible long-term threats of AI are a distraction from what’s being perpetrated on marginalized populations today, and I think if we approach technology ethically from the ground up, we may help people now and evade long-term harms.”
Always a Scientist in the Making
As a child growing up in Newton, Massachusetts, Dr. Rotemberg was an eager explorer of nature. “I had a very outdoorsy childhood — I always enjoyed hiking and being outside.” Dr. Rotemberg believes a love of the natural world is the first step in developing a love for science. “Being outside prompts a lot of questions. It inspires curiosity about new things and figuring out how things work.”
She never went through a fantasy career phase as a child — she didn’t dream of being a ballerina or a circus clown. She always knew that her future self would be wearing a lab coat.
“One of my first memories of someone asking me what I wanted to be when I grew up is of me telling them I wanted to be a biomedical engineer,” she recalls. She also remembers the surprised adult laughing in response.
Biomedical engineering is not a widely understood field — possibly because, as Dr. Rotemberg says, it is incredibly broad. “There are so many people studying so many different things who are all biomedical engineers,” she says. “In the most general sense, it’s a field in which you use or research engineering concepts in the context of biology or medicine.” For someone like herself, with countless interests, biomedical engineering was the perfect fit. “I think I was casting around for something that used all the things I liked,” she says.
She went to college at Harvard, where she managed to narrow her enthusiasms enough to select a major — physics. After graduating, she attended Duke University, where she entered an MD/PhD program. She earned her PhD in biomedical engineering, focusing on the non-linear behavior of tissue and how it can be monitored using ultrasound.
Dr. Rotemberg’s interest in imaging drew her to dermatology, where she looked forward to having continuity of care with patients and where she could study imaging in an accessible way. She was also inspired by a friend.
“My college roommate also went to med school and was already in her dermatology residency when I was still deciding on a specialty,” she says. “She convinced me to do a rotation in dermatology and I loved it. And now I see my friend at conferences every year!”
Amazed by Colleagues, Guided by Mentors
Because Dr. Rotemberg’s work is so interdisciplinary, there are several people in her lab whose areas of expertise are very different from her own. She delights in the “new, fascinating things” that come from her colleagues’ ideas and research that she never would have arrived at on her own. “I’m in awe of how different perspectives guide what we can accomplish,” she says.
As the head of her own lab, Dr. Rotemberg strives to share some of the valuable lessons she learned from her own mentors earlier in her career.
“If you’re lucky, your career will be long,” she recalls being told. “You’re sowing seeds today. You don’t know when they’re going to sprout, so you must have patience with yourself and the research.”
They are words of advice she has taken to heart. “There’s so much pressure to accomplish all the science today and stake your claim. I feel that pressure too, but I want to have a long career, and part of that means not burning out tomorrow, but thinking and investing in the long-term science.”
Another benefit of a long career is the extraordinary opportunity to learn.
“I’m constantly improving my scientific skills, my writing skills, my mentorship skills, and my management skills,” Dr. Rotemberg say. “I think of myself as continuously learning and developing into a better leader and manager, and a more well-rounded and insightful scientist.”
She admits to occasional frustration at a challenge common to many of her colleagues: Too little time. “I wish there were 500 more hours in every week,” she laughs, always mindful of the ultimate purpose of those hours in the clinic and the lab.
“I love that the work we’re doing has a real impact,” Dr. Rotemberg says. “We’re looking at questions that will really benefit patients, which is amazing. I feel fortunate to be at that intersection of studying complex phenomena and potentially leading to discoveries that will help patients and reduce harms.”