Ew Techniques for Imaging, Digitization and Analysis of Hree-dimensional Neural Morphology on Multiple Cales

bstract—Cognitive impairment in normal aging and neuroegenerative diseases is accompanied by altered morpholoies on multiple scales. Understanding of the role of these tructural changes in producing functional deficits in brain ging and neuropsychiatric disorders requires accurate hree-dimensional representations of neuronal morphology, nd realistic biophysical modeling that can directly relate tructural changes to altered neuronal firing patterns. To date owever, tools capable of resolving, digitizing and analyzing euronal morphology on both local and global scales, and ith sufficient throughput and automation, have been lackng. The precision of existing image analysis-based morphoetric tools is restricted at the finest scales, where resolution f fine dendritic features and spine geometry is limited by the keletonization methods used, and by quantization errors rising from insufficient imaging resolution. We are developng techniques for imaging, reconstruction and analysis of euronal morphology that capture both local and global tructural variation. To minimize quantization error and evalate more precisely the fine geometry of dendrites and pines, we introduce a new shape analysis technique, the ayburst sampling algorithm that uses the original grayscale ata rather than the segmented images for precise, continuus radius estimation, and multidirectional radius sampling