Computer-Aided Decoding of Brain-Immune Interactions in Gulf War Illness (GWI): A Joint Embedding on Brain Connectomic and Immunogenomic Markers
Technical Report,01 Sep 2017,31 Aug 2018
Trustees of Boston University Boston United States
Pagination or Media Count:
In this first year grant period, we developed an image processing pipeline for multi-parametric diffusion magnetic resonance imaging data, which has been collected from the Gulf War Illness Consortium. The pipeline was designed to apply the most up-to-date diffusion reconstruction techniques to provide multi-compartmental diffusion model based measures, q-space reconstructions as well as diffusion tensor metrics. As we planned and mentioned in the statement of work, major milestone in the year 1 of this grant was performing diffusion imaging feature extractions from the image data. As such, we processed data from 73 veterans for extracting major white matter bundles, a basis for cortical network reconstructions. In addition, we added gray matter diffusivity mapping to investigate potential microstructure changes in the cortex. Since the other milestone for year 1 to the middle of the year 2 period was building unimodal classifier of GWI neuroimaging and neuroimmune data, we started building unimodal classifiers on each features. Most significant findings in this project period was that the best classification results were observed in a classifier based on the advanced diffusion imaging measures. This may indicate that the advanced diffusion imaging measures might describe the underlying pathophysiology of GWI better than other markers. However, this work needs to be continued to the year 2 period to have clearer figures.
- Genetic Engineering and Molecular Biology
- Medicine and Medical Research