Skip Navigation
Search

Arie Kaufman, PHD

Distinguished Professor & Chief Scientist, CEWIT
Stony Brook University

Arie Kaufman is Distinguished Professor of Computer Science, Director of the Center of Visual Computing (CVC), Chief Scientist of the Center of Excellence in Wireless and Information Technology (CEWIT), and Site Director of the NSF Industry University Cooperative Research Center (IUCRC) for Visual and Decision Informatics (CVDI) at Stony Brook University. He served as Chairman of the Computer Science Department 1999-2017. He has been conducting research for 40 years in visualization, virtual-reality, medical imaging, machine learning and their applications, has published more than 350 refereed manuscripts, has delivered more than 20 invited keynote talks, has been awarded/filed more than 100 patents, and has been a principal/co-principal investigator on more than 130 researchgrants. He is a Fellow of the National Academy of Inventors (NAI), Fellow of IEEE, Fellow of ACM, was elected to the European Academy of Sciences, a recipient of the IEEE Visualization Career Award, and was inducted into the Long Island Technology Hall of Fame and the IEEE Visualization Academy, and the recipient of numerous other awards. He was the founding Editor-in-Chief of the IEEE Transaction on Visualization and Computer Graphics (TVCG), 1995-1998. He has been the co-founder/papers co-chair of IEEE Visualization Conferences, Volume Graphics Workshops, Eurographics/SIGGRAPH Graphics Hardware Workshops, and ACM Volume Visualization Symposia. He served as Chair and Director of IEEE CS Technical Committee on Visualization and Graphics. He received a PhD in Computer Science from the Ben-Gurion University, Israel, in 1977.

ABSTRACT

Reconstruction and Visualization of Neuronal Structures using Deep Learning

The understanding of neuron morphologies that underline brain function is central to neurobiology research. Recent advances in micro- and nano-resolution microscopy technologies, coupled with powerful machine learning techniques can enable neuroscientists to gain novel insights into the complex neural connections maps, thus leading to breakthrough understanding of human brain diseases. In this work, we present a novel end-to-end framework that can segment, reconstruct, predict, and visualize changes in neural fiber morphologies of biological specimens, imaged using micrometer resolution optical microscopes. Due to its inherent design, images from optical microscopes suffer from degraded contrast between the foreground and background, making it difficult to apply existing rendering techniques to effectively visualize the intricate neuronal structures. To this end, we have developed 2D and 3D convolutional neural networks to segment neurites using only a limited number of annotated ground-truth datasets. Using the results of our segmentation model, we can visualize weak neurites that were not visible in most neurobiology visualization frameworks. Moreover, the technical limitations of carrying out biological experiments preclude specimens from being sampled at more than one age time-point. We provide neuroscientists with the unique ability to predict changes in the structure of neuronal fibers across age time-points. This work is currently being used by our neurobiology collaborators to study the neurodegeneration in mice brains affected by Alzheimer’s Disease.