Johns Hopkins University
Towards Automatic Analysis of Multi-Terabyte Cleared Brains

Big imaging data is becoming more prominent in brain sciences across spatiotemporal scales and phylogenies. We have developed a computational ecosystem that enables storage, visualization, and analysis of these data in the cloud, thus far spanning 20+ publications and 100+ terabytes of public data including nanoscale ultrastructure, microscale synaptogenetic diversity, and mesoscale whole brain connectivity. In this project, we are extending our infrastructure specifically to be optimized to support cleared brain data, such as from CLARITY or iDISCO.  

Our first achievement for this project is the development of ndreg, a Python package ( that aligns cleared light sheet fluorescence microscopy images to atlases such as the serial two-photon Allen Reference Atlas. The ndreg pipeline aligns the data even with typical inhomogeneities, via the following steps: N4ITK bias correction,  SimpleITK’s affine registration and adaptive histogram equalization on both images, and finally a multiscale Large Deformation Diffeomorphic Approach to Registration using mean-squared error loss.  The result is improved registration over previous state-of-the-art, requiring under an hour of computation on a typical work-station. In subsequent years, we will integrate this stage of processing with cell detection, tractography, and quantitative neuroanatomy methods. All tools are developed open-source with suitable documentation to be useful for the community.

Key Research Resources Being Developed and Disseminated

  • Under development
    • Bloby: Cell detection software package that automatically identifies cell centroids based on intensity in the image volumes
    • Ndtractography: Tractography software package that automatically identifies axons/axon bundles using an intensity-based method


Joshua Vogelstein
Principal Investigator
Randal Burns
Co-Principal Investigator
Johns Hopkins University
3400 N Charles St
Baltimore, MD 21218-2688