Using Unsupervised Dimensionality Reduction to Identify Lesion Patterns Predictive of Post-Stroke Aphasia Severity
Interview with CBR Undergraduate Student Emerson Kropp about his recent research: Using Unsupervised Dimensionality Reduction to Identify Lesion Patterns Predictive of Post-Stroke Aphasia Severity
Introduction
Emerson Kropp is an undergraduate student at Boston University studying under Dr. Kiran in the Center for Brain Recovery as he completes his premedical bachelors degree in Health Science/ Human Physiology. He is also a certified EMT and alongside his work with the CBR he has gained experience with hands-on patient focused care and research.
Kropp’s research is focused on the impacts of lesion volume and location in stroke patients with aphasia. He explains, “the goal of this project is to apply non-negative matrix factorization (NMF) to data on brain regions damaged in strokes causing aphasia. In this way, NMF identifies large-scale stroke patterns which are represented across our patients, and some of these representations are much more correlated with aphasia and language ability than others.”
About the Research
In the CBR, Kropp worked with other members and alumni to collaborate and receive guidance on the project. PhD student Nicole Carvahlo and PhD recipients Isaac Falconer and Anne Billot notably helped to collect patient behavioral and MRI data for the MD project which provided a larger dataset that was able to be used in analysis for this project. He explained that, “Dr. Billot also helped create previous methods to calculate the percentage of spared gray and white matter based on patient lesions” which was used in this project’s analysis. Founding director of the CBR, Swathi Kiran, also provided guidance and feedback during the research process alongside Assistant Scientific Director Maria Varkanitsa.
Inspiration and References
The project took a lot of inspiration from the work of Dr. Anna Bonkhoff, Resident Physician at Massachusetts General Hospital and Harvard Medical School, and Dr. Julius Kernbach, Resident Physician in Neuroradiology at the University Hospital Heidelberg (UKHD). Kropp was inspired by their previous work1,2,3 applying non-negative matrix factorization (NMF) to stroke lesion data and correlating with behavioral measures. He explains, “my methods build off their papers in an attempt to further explore the usefulness of NMF in aphasia research, with a specific interest in common stroke patterns indicative of aphasia severity. The existing body of work surrounding voxel lesion symptom mapping and similar techniques has also been very important in providing a background framework for this project, as this provides useful comparisons for my results in terms of brain regions important in language function and recovery.” This previous work served as an informative starting point to launch Kropp’s study and help to explore new areas of research.
Looking Ahead
This project was also recently accepted to the Society for Neurobiology of Language. Kropp explains, “my hope is that this technique may provide insight into specific topographical stroke patterns significant for language dysfunction which could benefit future targeted treatment efforts.”
Kropp is looking forward to pursuing a career in neurology and his time with the CBR is helping him pursue his research interests alongside a team of passionate students and professionals in the field. As a member of the CBR team, he has led workshops on how to create lesion masks on neuroimaging data using ITKSnap, performed research under the guidance of professionals in the lab alongside his peers, and assisted in multiple projects in the lab. This coming fall, Kropp will be assisting core faculty member Andreas Charimidou and working on projects funded by the Alzheimer’s Association.
Referenced and related studies to this project
- Bonkhoff, A. K., Lim, J., Bae, H., Weaver, N. A., Kuijf, H. J., Biesbroek, J. M., Rost, N. S., & Bzdok, D. (2021). Generative lesion pattern decomposition of cognitive impairment after stroke. Brain Communications, 3(2), fcab110. https://doi.org/10.1093/braincomms/fcab110
- Bonkhoff, A. K., Hong, S., Bretzner, M., Schirmer, M. D., Regenhardt, R. W., Arsava, E. M., Donahue, K., Nardin, M., Dalca, A., Giese, A. K., Etherton, M. R., Hancock, B. L., Mocking, S. J. T., McIntosh, E., Attia, J., Benavente, O., Cole, J. W., Donatti, A., Griessenauer, C., Heitsch, L., … Rost, N. S. (2022). Association of Stroke Lesion Pattern and White Matter Hyperintensity Burden With Stroke Severity and Outcome. Neurology, 99(13), e1364–e1379. https://doi.org/10.1212/WNL.0000000000200926
- Kernbach, J. M., Hartwigsen, G., Lim, J., Bae, H., Yu, K., Schlaug, G., Bonkhoff, A., Rost, N. S., & Bzdok, D. (2023). Bayesian stroke modeling details sex biases in the white matter substrates of aphasia. Communications Biology, 6(1), 354. https://doi.org/10.1038/s42003-023-04733-1