A day in the life
Planner spread for the first week of August
My name is Claire Weissman and I am a second-semester senior Computer Science Major! This summer I got the opportunity to continue the research I did last summer with DePaul University. I am working in a group that specializes in machine learning on medical imaging. I am going to take you through what a normal day looks like for me! Half of my mornings start by waking up early so I can go to a 6am jiu jitsu class. After going to class, I come back and begin to get ready for my workday to start. This summer I am working an internship at a health care company as well as doing research with DePaul University which is what the Whitman Internship Grant is supporting. Once I am ready, I make my coffee or tea (depends on my mood) and set up my computers (work and school) at my desk and write down what I need to do that day. Then I check through my emails and deal with what needs to be addressed right away. For my research with DePaul most of the work I am doing is coding and writing the research paper for the Journal of Medical Imaging we are submitting at the end of August. I will break for lunch around 12 and chat with my roommates while I make something to eat, I also usually use this time to take a walk and get some fresh air. Then I work for the rest of the afternoon. Once my workday is done I usually drive about 20 minutes to the beach to end my day with a beach run.
This summer my tasks began with choosing a new dataset of medical images to use to test our method of calculating uncertainty per patient image. We created this method of
From left to right Lilly Roelofs, Roselyne Tchoua and me (Claire Weissman). Lilly is my partner in this research and Roselyne is our mentor and a professor at DePaul University
determining uncertainty last summer in order to help reduce the workload of radiologists’ by flagging images that are harder to diagnose and need more expert input. But after creating this metric and testing it on one data set that allowed for validation of our results, we wanted to reproduce the results with another dataset to show the generalizability of our metric. In order to start on this process, I spent a few weeks researching the available datasets of medical images on the internet that are open source. Many of the datasets that I found were either not complete or didn’t have the actual medical images but just other data from the images. Finally, we found a dataset that not only had images for each patient but also had radiologist annotations per patient in order to use as labels. This dataset was also very well documented and had already been used for some analysis, so it was also more trusted. Once the dataset was chosen, I spent around a month just organizing and cleaning up the data so that it could be run through the network properly.
Diagram from journal article I am working on. It shows the layout of the neural network being used in our research
After running the images through the network, I was also responsible for tuning the network such that the highest accuracy can be achieved. Once tuning the network was done I turned the data over to my partner Lilly who is working on creating the final results. Now I am just working on the final draft of our journal article so most of my days look like a lot of writing, taking breaks and then editing. I am also trying some other data analysis avenues to see if I can find any correlation between clinical data from the patients and our calculated uncertainty scores. I haven’t found anything worth writing about yet but I am still working on it!
Cool graph from my research paper
My work from home partner. Her name is Olive!
Experiences like Claire Weissman’s are made possible by the Whitman Internship Grant, which provides funding for students to participate in unpaid internships at nonprofit, some for-profit, and government organizations. We are happy to be sharing blog posts from students who were supported by either a summer, fall, spring, or year-long grants at organizations, businesses, and research labs all around the world. To learn how you could secure a Whitman Internship Grant or host a Whitman intern at your organization, contact us at ccec_info@whitman.edu.