Gibbs Assistant Professor, Yale University
Local diffusion geometry for calcium imaging analysis
Recent advances in experimental methods in neuroscience enable to record in-vivo large populations of neurons at cellular level resolution. The analysis of these complex datasets relies on extracting neurons as high-resolution regions of interest, while addressing demixing of overlapping spatial components and denoising of the temporal signal of each neuron. We propose a data-driven solution to these challenges, by representing the spatio-temporal volume as two graphs: one in the image plane and the other in the temporal domain. Our approach, based on global and local spectral analysis of these graphs, performs joint denoising and automatic extraction of structures from dense fluorescent images. To this end, we present a new greedy selective spectral clustering method capable of handling overlapping clusters and disregarding clutter. We demonstrate our approach on in-vivo calcium imaging of neurons and apical dendrites.
Gal Mishne is a Gibbs Assistant Professor at the Mathematics department at Yale University, working with Prof. Ronald Coifman. She received her B.Sc. degree summa cum laude in Electrical Engineering and Physics from the Technion-IIT, Israel, in 2009, and her Ph.D. degree in electrical engineering from the Technion in 2017.
Gal’s PhD, under the supervision of Prof. Israel Cohen, focused on novel manifold learning and deep learning methods for anomaly and target detection in high-dimensional data. In her post-doc research, she has been developing a data-driven and model-free approach for the analysis of calcium imaging of neuronal activity in awake behaving mice. Her research addresses three aspects of analyzing neuronal and behavioral activity, from low-level processing through mid-level data organization to high-level inference and prediction.
Her research interests include manifold learning, applied harmonic analysis, image processing and biomedical signal processing.