U01 - Fischl 1U01MH117023-01

Imaging and analysis Techniques to Construct a Cell Census Atlas of the Human Brain

Our consortium will develop and utilize an imaging infrastructure to create a human brain cell census and instantiate it in a coordinate system that will enable an immediate impact of all in vivo MRI studies of the human brain. Our imaging pipeline begins with 150 μm isotropic ex vivo structural MRIs and 500 μm 90 direction diffusion MRI. 4x4x2 cm tissue blocks will then be imaged with serial sectioning polarization sensitive optical coherence tomography (PS-OCT) to obtain 20 μm isotropic resolution images of cyto- and myelo-architectural features and fiber tractography. Registration with the MRI is preserved because imaging is performed before sectioning. Serial sections with 150 μm thickness are cut from the block face, cleared, and subsequently imaged by selective plane imaging microscopy (SPIM) with 1 μm isotropic resolution, and with multiplexed histological staining of multiple molecular markers. Segmentation of common features amongst the SPIM, OCT, and MRI, permits registration of the cellular information from SPIM back to the MRI coordinate system with OCT serving as an important intermediary permitting registration of SPIM with MRI with micrometer resolution. Our aims are designed to establish our imaging workflow and build a cell census atlas for language regions in human frontal cortex, resulting in a clear path forward for obtaining the cell census for the entire human brain. 

Project Leadership

Bruce Fischl, Ph.D. (Principal Investigator) 
Professor in Radiology, Harvard Medical School
Neuroscientist, Massachusetts General Hospital
Director, Computational Core, Martinos Center, MGH


David Boas, Ph.D. (Principal Investigator)
Professor, Department of Biomedical Engineering, Boston University
Director, Neurophotonics Center


Patrick R. Hof, M.D. (Co-Investigator)
Regenstreif Professor and Vice Chair
Fishberg Department of Neuroscience
Hess Center for Science and Medicine


Francesco Pavone, Ph.D. (Co-Investigator)
Professor, Department of Physics, University of Florence 
European Laboratory for Non Linear Spectroscopy (LENS)

Project Data Types

  • We will acquire structural and diffusion MRI data on post-mortem human hemispheres and process the data using an adapted version of FreeSurfer to generate cortical surface models, architectonic boundary predictions, laminar models and subcortical segmentations.
  • We will acquire 3-m isotropic OCT data of 4x4x2 cm blocks of tissue from the frontal lobes of the previously mentioned hemispheres. During OCT acquisition, the tissue will be cut into 150 μm-thick sections, which will be sent to Dr. Pavone’s lab for further analysis.
  • The 150 μm sections will be cleared and imaged with multiple fluorescent stains to generate the molecular information needed by Dr. Hof to develop a cellular typology.
  • Dr. Hof’s and Pavone’s labs will manually classify a training set of cells that will be used to train a CNN to automate cellular classification in the full samples.

Related Resources

  • We will provide tools for mapping histological information to in vivo studies using the surface-based analysis implemented in the FreeSurfer suite of neuroimaging analysis algorithms: https://surfer.nmr.mgh.harvard.edu/fswiki