Directional multiscale representations for phenotypic profiling in neuronal imaging.
Medical conditions, such as neurodegeneration and neuropsychiatric disorders, are preceded in preclinical stages by subtle alterations in neuronal morphology and intracellular distribution of key molecular components (visualized through fluorescent microscopy) that conventional data analysis tools are inadequate to detect and quantify. There is a compelling need to go beyond these data analysis limitations that hamper the modelling of critical biological processes preceding the diseased state. In this talk, I will present an innovative theoretical framework based on directional multiscale representations to capture complex morphological structures in multidimensional data in a highly sensitive and comprehensive manner. I will show the application of these ideas for the profiling of multiple phenotypic characteristics in 3D high-resolution fluorescent images of neuronal cultures and brain tissue. This method will facilitate the development of a high-throughput quantitative platform for the study of neuronal networks and local brain circuits.