CSpotlight: Experiencing research as an undergrad
Gaoxiang Luo always knew he wanted to conduct research at the University of Minnesota. The pandemic inspired this B.S. student to apply his knowledge in deep learning to the healthcare field. Thanks to his enthusiasm and dedication, he was able to be a part of the groundbreaking research team that developed AI algorithms to analyze chest X-rays for COVID-19.
Why did you choose to study computer science at the University of Minnesota?
I believe that computer science helps people better understand the world. Almost all fields of study need computer scientists to be successful. For instance, last year DeepMind AI made gigantic leap in solving a 50-year-old grand challenge in the field of protein structure, which was completely amazing to me.
I knew that I wanted to do academic research during my college years, and the University of Minnesota, as a public land-grant research institution, offered so many research opportunities for undergrads, as well as wonderful faculty members happy to mentor undergrads.
What sparked your interest in computer science? Specifically, how did you become interested in medical imaging?
My interest in medical AI started at the beginning of the global coronavirus pandemic. I was participated in a COVID-19 forecasting project held by the University's SIAM (Society for Industrial and Applied Mathematics) chapter and Ecolab. This was my first time hearing about neural networks, and I was amazed by how powerful they could be. That summer, I came across a talk by professor Ju Sun and learned about diagnosing COVID-19 based on chest X-rays. I emailed him and expressed my interest in medical imaging analysis, and was able to join his research team. In the meantime, I had started to take machine learning classes through Coursera in my free time, which allowed me to take the special topics course Think Deep Learning in the fall of 2020.
Tell us more about your research projects with professor Ju Sun and professor Shashi Shekhar.
At first, I was working with assistant professor Ju Sun on developing AI algorithms to analyze chest X-rays for COVID-19. Now, I am co-leading related research on rib fracture detection, in collaboration with M Health Fairview and under the supervision of Professor Sun. With a mortality rate of 1 in 5 for patients over 65 years old, rib fractures are one of the most common sources of morbidity and mortality in older patients following trauma. The clinical decision support system that we are building will assist radiologists to diagnose whether a patient has rib fractures (classification) from a chest x-ray and where they are (detection/localization) in CT scans to assist radiologists. The classification aspect will soon undergo testing for internal clinical use. We are still waiting on the annotated CT-scans dataset. Since obtaining medical data is expensive, we’re also working on Federated Learning to train a global AI model collaboratively with other universities without sharing local data.
Meanwhile, I've been working on applying spatial variability aware neural network (SVANN) to map wetlands in Professor Shashi Shekhar and his spatial computing group. I actually had the chance to present my work during the 2021 Virtual Undergraduate Research Symposium.
How did you become involved with the new Medicine and Machine Learning (MaML) student group?
This group just started meeting in the fall of 2020. I particularly enjoy this group because the focus perfectly aligns with my research interests. I've been especially inspired by the group's podcast series featuring interviews with experts in AI and healthcare. Over winter break, I had the opportunity to participate in the first-ever MaML Coursera Dash, where we took the AI for Medicine courses with a buddy in the group. It was great that we could tackle these courses together, and keep each other accountable and motivated.
Tell us more about your position as a student reviewer for the Minnesota Undergraduate Research & Academic Journal (MURAJ).
I serve as a computer science and mathematics reviewer at MURAJ. Originally, I was interested in getting exposed to peer-reviewed papers, but the experience went way beyond what I expected. I was able to receive systematic training, designed by the wonderful MURA editorial board, that provided a basic understanding of peer review. I also got to attend talks given by university professors and librarians, where they shared more tips on reviewing academic papers. I was able to author the first computer science feature for the MURAJ: In Focus publication. My article, Application of Artificial Intelligence to Help Fight COVID-19 was published in March 2021.
What advice do you have for incoming computer science students?
If you’re interested in research, there are many resources available to you. To start, I'd recommend browsing professors’ websites and learn about their current research. You might even want to email them and ask to audit their weekly group meeting. I actually joined two research groups this way! You can always ask them about the prerequisites to join their labs so you know how to prepare yourself. The Office of Undergraduate Research also has many valuable resources and the Computer Science & Engineering department sometimes offers an introduction to undergraduate research course (CSCI 2980).
What are your plans after graduation?
Besides research, I have really enjoyed sharing my expertise and tutoring students as a teaching assistant within the CS&E department and as a tutor at SMART Learning Commons (a part of the University of Minnesota Libraries). So, I plan to pursue a Ph.D. degree in computer science to be a researcher focused on building trustworthy and fair AI. At this point, I am planning to stay in academia after completing my graduate studies.