Data Driven Interference In Diffusion MRI: Deep Learning Harmonization of Quantitative Brain Biomarkers
White matter changes are increasingly implicated in early Alzheimer's Disease progression, and diffusion weighted magnetic resonance imaging
(DW-MRI) has been included in many national-scale studies. Yet, quantitative investigation of DW-MRI data is hindered by a lack of consistency due to variation in acquisition protocols, sites, and scanners. DW-MRI enables quantification of brain microstructure and facilitates structural connectivity mapping. Substantial recent progress has been made with calibration and harmonization to reduce inter-subject variance and improve interpretability of computed measures. Yet, the fundamental challenge remains that clinical application of DW-MRI (as currently implemented) is confounded by inter-scanner and inter-site effects. With the wide availability of imaging data, the field would appear to be ripe for machine learning approaches. Indeed, recent advanced in image-based deep learning have dramatically reduced distortions associated with rapid imaging. On a regional basis, deep learning has shown the ability to capture individually specific white matter tracts using context tuned blocks, while voxelwise deep models have shown leading agreement with external histological validation.
Despite these successes, challenges remain in terms of robustness, explainability, and usability for deep learning models to gain widespread acceptance. In this talk, we will discuss successful applications of deep learning with diffusion MRI and opportunities for deep learning innovation with diffusion MRI data.
Bennett A. Landman, Ph.D. is Professor and Department Chair of Electrical and Computer Engineering at Vanderbilt University, with appointments in Computer Science, Biomedical Engineering, Radiology and Radiological Sciences, Psychiatry and Behavioral Sciences, Biomedical Informatics, and Neurology. He graduated with a bachelor of science
(’01) and master of engineering (’02) in electrical engineering and computer science from the Massachusetts Institute of Technology, Cambridge, MA. After graduation, he worked in an image processing startup company and a private medical imaging research firm before returning for a doctorate in biomedical engineering (‘08) from Johns Hopkins University School of Medicine, Baltimore, MD. From 2010 to 2021, he severed on the Faculty of the Electrical Engineering and Computer Science Department, Vanderbilt University, Nashville, TN. In July 2021, he joined and became the first chair of the newly formed Electrical and Computer Engineering Department. His research concentrates on applying image-processing technologies to leverage large-scale imaging studies to improve understanding of individual anatomy and personalize medicine.Dr.
Landman has received grant funding from the National Institutes of Health, the National Science Foundation, the Department of Defense, and industry support. He is highly collaborative with 340+ co-authors across disciplines, career stages, and institutions, resulting in 340+ peer-reviewed publications and 9,500+ citations. He served on the MICCAI Society Challenge Working Group, as co-chair of the SPIE Medical Imaging Image Processing conference (2017-2021), as co-chair of the SIIM Machine Learning Tools Committee (2018-2021), and on the editorial boards of the IEEE Transactions of Medical Imaging (2015-) and SIIM Journal of Digital Imaging. He has organized 11 workshops and challenges at MICCAI since
2011 and has supported challenges with SPIE, ISBI, ISMRM, and Kaggle. He served founding director of the Center for Computational Imaging at the Vanderbilt University Institute of Image Science and as chair of the faculty advisory board of the Vanderbilt University Advanced Computing Center for Research and Education (ACCRE). He is currently the Principal Scientist of ImageVU, Vanderbilt’s clinical data reuse initiative in Radiology.