Faster, Deeper, and Free: Opening the Door to Better Biomedical Imaging through Signal Processing and Machine Learning

Signal processing and machine learning are expanding the capabilities of biomedical imaging modalities such as magnetic resonance imaging and optical microscopy in many new directions. To begin, we revisit deep learning solutions to the image reconstruction problem in cardiac magnetic resonance imaging. In this work, we mitigate image quality variability in clinical patient data used for training deep artificial neural networks. We also improve the accuracy and precision of highly undersampled acquisitions for quantitative imaging in such settings. Next, we turn our attention to image processing and analysis for optical microscopy of the murine brain. We describe how we are automating and increasing the robustness of image enhancement and segmentation techniques under development for our forthcoming open-source Neuroglia Image Toolkit, to make them suitable for widespread use. We also describe how we reconstruct and label large brain regions at single-cell resolution. These techniques promise to transform imaging-based approaches for behavioral neuroscience and for understanding neurological disorders such as epilepsy and Alzheimer's disease. These projects are collaborations with Drs. Scott Acton, Ukpong Eyo, Jaideep Kapur, Christopher Kramer, Gustavo Rohde, Michael Salerno, and Cedric Williams, all at the University of Virginia.
Daniel Weller joined the faculty of the University of Virginia in August 2014, after completing a two-year postdoctoral fellowship at the University of Michigan, supervised by Jeffrey Fessler and Douglas Noll. He completed the PhD and SM (Master's of Science) in Electrical Engineering in 2012 and 2008, respectively, both at the Massachusetts Institute of Technology, supervised by Vivek Goyal. Before that, he completed the BS in Electrical and Computer Engineering in 2006 at Carnegie Mellon University, with university and college honors. His research interests are in signal processing and biomedical imaging, including magnetic resonance imaging and optical microscopy. He is an associate editor of the IEEE Transactions on Medical Imaging and a member of the Computational Imaging Technical Committee of the IEEE Signal Processing Society, and he served on the organizing committee for the 2018 IEEE International Symposium on Biomedical Imaging (ISBI). His research has been sponsored by the National Science Foundation, the National Institutes of Health, the Thomas F. and Kate Miller Jeffress Memorial Trust, Bank of America, Trustee, and NVIDIA Corporation (provided GPU's used in this research).