Research Internship: Ethnoracial Bias in AI Methods for Brain MRI Analysis

Research Internship: Ethnoracial Bias in AI Methods for Brain MRI Analysis
Dina Zemlyanker and Prof. Joan Eugenio Iglesias, Laboratory for Ex Vivo Modeling of Neuroanatomy (LEMoN), Martinos Center for Biomedical Imaging, Harvard Medical School
Fall 2024 and beyond

The LEMoN (lemon.martinos.org) is looking for an undergraduate research assistant (RA) to investigate ethnoracial biases in artificial intelligence (AI) methods for automated analysis of brain MRI scans. Our group has developed publicly available AI tools that can analyze large, uncurated collections of clinical scans. However, it is still unclear how the performance of our tools (and methods of other labs) depend on the race and ethnicity of the analyzed subjects. To investigate such biases, we have recently downloaded a large dataset with over 100,000 scans from Black, White, and Asian subjects, which we plan to use for a wide spectrum of AI bias studies.  Responsibilities: This RA will work with researchers at the LEMoN to (i) help organize the dataset; (ii) run AI methods on the data; (iii) organize and analyze the results; and (iv) investigate bias mitigation methods. Requirements and Expectations: The RA is expected to have some basic knowledge of Linux and statistics and will have plenty of opportunities to develop their AI and Python skills via interaction with other LEMoN members, participation in group meetings, and the LEMoN reading group. Additional Information: Start date and hours/week are flexible, but we expect the student to stay at least two terms – particularly if the hours/week are low. Location: Charlestown Navy Yard (hybrid/remote possible). To Apply: please send a brief cover letter and CV to Dina Zemlyanker at  dzemlyanker@mgh.harvard.edu. (posted 8/2024)