Practicum Pride: UCSF Neuroscape
Jenny Kong came to USF’s MSDS program after obtaining a Bachelor’s Degree in Economics and working as a Data Analyst in Taiwan. Continue reading to learn about her work at the neuroscience center at UCSF!
Can you tell us a bit about UCSF Neuroscape? What is it like working there?
The UCSF Neuroscape Lab is a lab under the UCSF Neuroscience Center. Working here is similar to most research settings : (extremely) detail-oriented, somewhat serious but somewhat laid-back in a sense. My practicum mentor is a psychology professor who would explain brain structures to me using oranges as an analogy (though still talks about Talairach coordinates and warping of those oranges) so I always feel encouraged to ask questions.
This project is part of the Wicklow AI in Medicine Research Initiative (WAMRI) at the Data Institute. Can you describe the project(s) that you are working on?
The goal is to use machine learning with fMRI data to classify network patterns of concurrently activating brain regions that arise during successful high-fidelity memory retrieval. The project scope includes applying machine learning algorithms to perform sensitivity-based feature selection and performing group analysis among young/older adults. We’re approaching the end of the project so we’re expecting to start the write-ups for publication in a couple of weeks.
How are you applying the knowledge gained from the program to your practicum? Is there a particular class that has been the most helpful?
I started working after taking a few weeks of introductory machine learning classes in our program and found the knowledge gained from class surprisingly useful. I still remember the first time I met my mentor — I thought it was just a welcoming meeting with a tour around the lab, but he started asking questions regarding application of kNN and SVMs on multi-voxel pattern analysis, and I calmly went through the whole discussion because I had just submitted a homework on the same topic. There are often scenarios where I have to explain the rationale behind model selection to my mentor. When that happens, I have to dig deeper into those algorithms beyond my class’ scope. My faculty mentor from USF also provides insightful advice which helps a lot.
What has been your experience working in a research setting? How does it compare to working in a traditional office setting?
It still seems unreal to me that I’m working on a research project at UCSF, in fact, this opportunity was actually the reason I applied to this program. I enjoy the tranquil academical atmosphere, and the fact that the purpose of our work is to better understand human brains and contribute to future research on activation of brains and amnestic dementia.
The most obvious difference between research centers and commercial firms is that we make rather slow but steady progress. When I worked as a data analyst, there were incoming sales data 24/7 and we had to adjust our forecast daily. It was fast-paced, involving international meetings and many small and prompt decisions. Here, every step moving forward is carefully examined and a consensus is reached before we make the move. Upon answering this question, it strikes me that we’re looking at fMRI data collected back in 2014.
What is the biggest challenge you’ve faced at your practicum?
Since we only have onsite meetings once a week, I work remotely most of the time. This is particularly challenging because of my lack of domain knowledge. I once spent hours debugging because I thought my code was not generating the right output, but when I showed it to my mentor he quickly identified that this was because the file was misnamed by someone and I was using the wrong data. Also, although we have Python libraries that read in the 4-D fMRI data into numpy arrays, it is seemly impossible for me to do a sanity check even after transforming them back to 4-D space. It requires good communication skills to always make sure that we’re on the right path, and understand what we’re predicting and the inferences behind those predictions.
Are there any cool perks?
Unfortunately I haven’t tried the VR games they designed for the participants to detect whether the activation of brain voxels changes before/after playing the games. However, my mentor did say that I could probably get my own structural MRI as a souvenir in the end…I hope he sees this article someday.
For more information on the MS Data Science program:
For more information about the Wicklow AI in Medicine Research Initiative (WAMRI): https://wamri.ai/
