DGHI20 Reflections: Mercy Asiedu PhD’19 and Chris Lam MS’12, PhD’18

Not long after Chris Lam, Ph.D., began his doctoral studies in biomedical engineering in 2012, his mentor, Duke Global Health Institute professor Nimmi Ramanujam, Ph.D., asked a question that changed everything. Ramanujam’s team had been working on a less invasive way to screen women for cervical cancer, but their prototype, which used light to detect signs of cancer in cervical tissue, wasn’t performing as expected. 

“Nimmi said, ‘Why can’t we use an inexpensive camera?’“ recalls Lam, who is now a technical lead for optical engineering at Masimo, a medical device company in Irvine, California. “And that really started us brainstorming about how to do that, and how to make it portable and low-cost.”

That inspiration launched the development of the pocket colposcope, one of the most promising and compelling innovations to arise during DGHI’s first 20 years. A slimmer, less-expensive alternative to traditional devices used in cervical examinations, the pocket colposcope is already helping make cervical cancer screening less painful and more accessible for women in low- and middle-income countries, where undetected cervical cancer cases still cause thousands of deaths each year. 

When Mercy Asiedu, Ph.D., joined Ramanujam’s team as a doctoral student in 2014, she led the design of a second potentially game-changing device called the Callascope, which eliminates the need for a speculum in cervical exams, opening the door to cervical cancer screening that women could do at home. She also developed software and AI models that integrated with the imaging devices. Asiedu, who is now a research scientist working on AI health applications for Google, co-founded the Calla Health Foundation to commercialize the innovations, which have now been used in clinical trials and pilot studies in 11 countries. 

As part of our DGHI Reflections series looking back at key moments in the institute’s first 20 years, we spoke with Asiedu and Lam about the many ups and downs in the years-long process of building and testing those devices, and what it taught them about designing health technologies that can work in resource-limited environments. Both were doctoral scholars with DGHI during their time in Ramanujam’s lab. Lam earned a Master of Science in Global Health from DGHI in 2012. Asiedu completed a graduate certificate in global health during her doctoral studies. The conversation has been edited for length and clarity.

What experiences did you each have before coming to Duke that made you interested in using engineering to solve health problems?

Lam: I did my undergraduate degree in biomedical engineering at the University of Cincinnati, and then I worked in clinical and basic sciences research for a few years before deciding to go back to graduate school. I was in the second cohort of the DGHI master’s program, and during that time, I was also a teaching assistant for [DGHI and biomedical engineering professor] Robert Malkin’s design class. I really liked his approach of applying biomedical engineering to global health, and that got me interested in pursuing this area as a doctoral student.   

Asiedu: I actually wasn’t originally interested in engineering, but that’s partially because I didn’t even know what biomedical engineering was. I wanted to be a doctor. I grew up in Ghana, and I was seeing these health disparities all the time. One of my favorite aunts died in childbirth, and it just seemed like people were constantly passing away for unnecessary health reasons. I thought becoming an OB/GYN would be the best path to making a difference. 

As an undergraduate, I decided to do biomedical engineering just as a safety net in case I didn’t go to medical school. But I really enjoyed the program. I think I realized that being able to build tools that could help patients could potentially be more scalable than being a doctor. 

Chris, I know you joined Nimmi Ramanujam’s lab first, back in 2012. Was the pocket colposcope even an idea at that point?

Lam: So when I joined the lab, I was taking over a project from a student who had just graduated. They were trying to use diffuse reflectance spectroscopy to interrogate suspected pre-cancerous and cancerous regions in the cervix in a noninvasive way, and they had tested a prototype in Haiti. I went there twice to collect more data, but there were a lot of things like the fragility of the sensor and the environmental sensitivities of the equipment that would preclude the successful implementation of that technology in a resource-constrained setting. So we were asking, do we keep pursuing this? And Nimmi said, “Why can’t we use an inexpensive camera?” And that really started us brainstorming about how to do that, and how to make it portable and low-cost. 

I found this spy-pen camera, a tiny camera on the tip of a pen with a little flash drive, and I took it apart. And I thought, could we put this camera on the tip of an endoscope or something to get closer to the cervix and image it? I remember someone brought in a bunch of different feminine hygiene products, and we just started looking at all of them to find a form that might work. It was just a lot of brainstorming about the technology and rapid prototyping. 

Mercy, where was the project when you joined the team?

Asiedu: When I joined, one of the early prototypes of the pocket colposcope was already in clinical studies, so I started out helping with clinical testing. And I think that’s one of the things Nimmi does really well – in your first few months, you really get to understand how your device will be used with patients. At the same time, I was starting to work on a design for a self-screening tool. During my interviews with Nimmi’s lab, I had met Chris, and he had said this was one of the goals of the team, to make something that could work without a speculum. And the idea of improving comfort – and self-screening – was really exciting to me. 

So the pocket colposcope prototypes still required use of a speculum to see the cervix, right?

Asiedu: Yes. One of the issues is that, when you put a tampon-like camera into the vaginal canal, the walls kind of close above the camera, and so you don’t actually see the cervix. So the pocket colposcope was being used in clinic with a speculum to expand the vagina, and my task was to figure out how to make it work without a speculum. And that was what led to the Callascope. 

What were some of the other challenges you ran into in developing the devices?

Lam: One of the things we had to think about was how do we clean the device between uses? Is there going to be a disposable sheath, or is it going to be something that you can immerse in a cleaning fluid? And would that cleaning solution be readily available in limited-resource settings? I had many failed projects where we tried different methods to make sure we could submerge the camera for disinfection. That was one of the design challenges we had to confront in thinking about how this was going to be used in the field. 

Another was how to run the imaging on the device. Initially, I had created this programming script that we were running on a computer to capture the images, but if we wanted a nurse to be able to do this in the field, we needed something easier to use. And so Mercy came up with this great graphical interface that could run on a tablet. But I’ll let Mercy talk about that …

Asiedu: Oh, I had forgotten about that. I had worked on user interfaces as an intern with the Mayo Clinic, and so when I joined the team, I thought it would be nice if doctors could just see buttons and images, so I coded up a computer-based interface. We got a computer science student to build the first Android version, and now we have an app that allows the camera to connect to a phone and save images to the cloud. That was important in making it something clinicians would be able to use.

Another thing that came from taking the colposcope to the field is that we saw that sometimes community health workers would have to wait a long time to consult with an OB/GYN to validate the images. The idea came up to provide some kind of automated assistance to the community health workers. And so I developed and evaluated machine learning algorithms that could help them interpret the images they were getting with the device, in collaboration with [former Duke computer science professor] Guillermo Sapiro.

Was there a particular moment where you felt like you were really on to something, that this approach could work?

Lam: I’d say it was when we started to see some of the images coming in from the field. Once we were able to get a beta prototype out to clinics in Tanzania and Peru, we started looking at the data, and the images we were getting could easily have come from a $20,000 colposcope. And that was the coolest thing, seeing that we could get that quality of images from our little device. 

Asiedu: By the time I joined, I think it was pretty clear the pocket colposcope was going to succeed. But for the Callascope, it just seemed like it was going to be a very hard task. It had to be small but also had to open up the area close to the cervix while withstanding a lot of pressure. I designed a lot of really scary-looking early prototypes, with a lot of very small moving parts. But what ended up working was actually a lot simpler, and I think the first time it worked, we were just really surprised we could do it with such a simple design. 

The moment that drove it home for me was when we were testing the device at a hospital in Ghana. We’d had a lot of feedback on our papers and grants saying that people in developing countries would never use a device like this, that it was too complicated. And at first, a lot of women were reluctant to try it. But then this one woman tried it, and she came out talking to others who were there about how easy it was. And then everybody wanted to try it. And I knew we were on to something. 

And now you’re both working on real-world applications of health technologies. Did your experiences on this project influence the direction you’ve taken in your careers?

Asiedu: Yeah, I think it had a huge influence. During my last year at Duke, an opportunity came up with the Schmidt Science Fellows program, which was looking for people with a lot of interdisciplinary experience. And I think a big reason why I was selected was because this project was so interdisciplinary and impactful. Through that program, I was able to spend three years working on health applications and additional machine learning research, and that led to the position I have now.

I’m currently working on machine learning–based digital biomarkers from wearables for women’s health and evaluating how large language models perform in different countries and across different languages. Those are areas I’m passionate about, but I have always tried to stay involved with health devices, and that is definitely because of the experience I had working with Nimmi.

Lam: For me, I think it was just the opportunity to do so many different things. With Nimmi’s team, I was kind of a jack of all trades. I was doing nitty-gritty mechanical design, circuit boards, image quality assessment and even a little bit of software. Now, my job is more about applying knowledge to refine R&D systems, but I think it was so valuable to have that broad exposure to the whole process. We were even helping with the FDA 510(k) submission for the pocket colposcope, which is something most graduate students do not get a chance to experience while in school. 

I think the other thing that was just so valuable is how we were able to take our designs into the field. It’s really rare for students in science and engineering to be able to do that, because the process can just take too long. But in Nimmi’s lab, we were prototyping and going to the clinic right away. We were trying to solve real-world problems and getting feedback from patients so we could improve our designs, and you can’t beat that kind of experience.