Advanced instrumentation and cellular imaging techniques using high-throughput 3D electron microscopy are driving a new revolution in the exploration of complex biological systems by providing near seamless views across multiple scales of resolution. These datasets provide the necessary breadth and depth to analyze multicellular, cellular, and subcelluar structure across large swathes of neural tissue. While these new imaging procedures are generating extremely large datasets of enormous value, the quantities are such that no single user or even laboratory team can possibly analyze the full content of their own imaging activities through traditional means. To address this challenge, we propose to further develop and refine a prototype hybrid system for high-throughput segmentation of large neuropil datasets that: 1) advances automatic algorithms for segmentation of cellular and sub-cellular structures using machine learning techniques;2) couples these techniques to a scalable and flexible process or tool suite allowing multiple users to simultaneously review, edit and curate the results of these automatic approaches;and, 3) builds a knowledge base of training data guiding and improving automated processing. This system will allow project scientists to select areas of interest, execute automatic segmentation algorithms, and distribute workload, curate data, and deposit final results into the Cell Centered Database (Martone et al. 2008) via accessible web-interfaces.

Public Health Relevance

Emerging techniques in cellular and subcellular 3D imaging are generating datasets of enormous value to the study of disease processes and to the pursuit of greater insight into the structure and function of the nervous system. To unlock the potential of these data, new solutions are needed to improve the capability to segment and label the individual molecular, subcellular and cellular components within very large volumetric expanses. To address this challenge, we propose a hybrid system for high-throughput segmentation of large neuropil datasets that advances machine learning algorithms for automatic segmentation and couples these techniques to a scalable tool suite for multiple users to simultaneously review, edit and curate results.

National Institute of Health (NIH)
National Institute of Neurological Disorders and Stroke (NINDS)
Research Project (R01)
Project #
Application #
Study Section
Special Emphasis Panel (ZRG1-IMST-L (90))
Program Officer
Liu, Yuan
Project Start
Project End
Budget Start
Budget End
Support Year
Fiscal Year
Total Cost
Indirect Cost
University of California San Diego
Schools of Medicine
La Jolla
United States
Zip Code
Liu, Ting; Jones, Cory; Seyedhosseini, Mojtaba et al. (2014) A modular hierarchical approach to 3D electron microscopy image segmentation. J Neurosci Methods 226:88-102
Giuly, Richard J; Kim, Keun-Young; Ellisman, Mark H (2013) DP2: Distributed 3D image segmentation using micro-labor workforce. Bioinformatics 29:1359-60
Jurrus, Elizabeth; Watanabe, Shigeki; Giuly, Richard J et al. (2013) Semi-automated neuron boundary detection and nonbranching process segmentation in electron microscopy images. Neuroinformatics 11:5-29
Seyedhosseini, Mojtaba; Tasdizen, Tolga (2013) Multi-class multi-scale series contextual model for image segmentation. IEEE Trans Image Process 22:4486-96
Seyedhosseini, Mojtaba; Kumar, Ritwik; Jurrus, Elizabeth et al. (2011) Detection of neuron membranes in electron microscopy images using multi-scale context and radon-like features. Med Image Comput Comput Assist Interv 14:670-7