|
|
|
Table of Contents What is a Sparse Synapse Resolution Brain Connectivity (SSRBC) Atlas? What Neuroanatomical Facts can be Derived Using an SSRBC Atlas? Links to the "Extreme Neuroanatomy" Research Community
35 Steps in the Creation and Use of a Single Brain Physical Slice Library (SBPSL) (SLIDE SHOW) What Types of Experiments can be Performed by Remote Researchers Using a SBPSL? Automated Taping Lathe-Microtome Prototype Development (SLIDE SHOW) Movies of Lathe Microtome cutting and tape collection in action! 20 Second *.AVI file (7 Mbytes) 3 Minute *.AVI file (55 Mbytes) Software Development (SLIDE SHOW) SBPSL Proposal Paper (PDF Document) SBPSL Full PowerPoint Presentation (Warning large file! *.ppt file is 29Mbytes) SpinalSeries7um.zip (12 *.bmp files) Dendritic Explorer test program overview slide
| What is a Sparse Synapse-Resolution Brain Connectivity Atlas?Defined on this page:
Today, there exist many brain atlases consisting of uniform slices of a single brain (either animal or human) cut in one of the cardinal directions (horizontal, coronal, or sagittal). These atlases are stained such that cell body densities or myelinated axon densities are visible when imaged with a video light microscope. Such atlases provide raw data for neuroanatomists, who use them to delineate the cytoarchitectonically identifiable regions and the major axonal pathways of the brain. Once delineated, the regions and pathways are usually annotated on a duplicate set of the atlas' images. Such atlases are extremely valuable to physiological recording experimentalists, fMRI interpreters, and to neuroanatomists themselves. The later uses the atlases while performing more detailed tract tracing experiments which require landmarks and uniform coordinate frames in order to be interpreted correctly. The highest resolution mammalian brain atlas today (probably the High-Resolution Mouse Brain Atlas produced by Richard L. Sidman at Harvard) uses 10 microns resolution images on 20 micron slices allowing individual cells to be made out when using the proper stain. This resolution, however, is totally insufficient to resolve individual axons or dendritic processes much less synapses. In short, such atlases can be used to delineate regions based on gross cytoarchitectonic differences, however, such atlases are completely useless (by themselves) in determining the connectivity of a brain. To determine such region-to-region connectivity, hundreds of in vivo tract tracing experiments are performed on separate animals, the animals sacrificed, and imaged brain slices of the tracer-injected brains are compared to a standard atlas to determine which regions project to (or receive projections from) a tracer injected region. Sidman (and others) have suggested the creation of a 1 micron resolution atlas (which would contain more than 1000 times the total number of voxels in the current High-Resolution Mouse Brain Atlas). Such an atlas would for the first time allow the direct tracing of the myelinated axons in a mouse brain. The inter-regional connectivity could be directly read off of such an atlas without the need for in vivo tract tracing experiments, and more importantly without the need to interpret and register data across different experiments and brains. Here we define such an atlas, whose resolution and staining protocols allows direct connectivity tracing, as a "Brain Connectivity Atlas". Synapse-Resolution Brain Connectivity Atlas ... Even at 1 micron resolution, the majority of dendritic processes and all synapses are invisible. Only a Transmission Electron Microscope (TEM) can reliably probe neural tissue and reveal the details of synaptic connectivity between neurons. A TEM providing a resolution of approximately 10 nanometers would be needed to reliably map the synaptic connectivity of a brain. This 10 nanometer voxel-resolution TEM atlas is what we define here as a "Synapse-Resolution Brain Connectivity Atlas". Such an atlas would represent an increase of a factor of 1 million in the total number of imaged voxels above even the ambitiously proposed 1 micron resolution axon tracing atlas described above. Such an incredible volume of imaging (and more importantly the difficulty of handling perhaps millions of extremely thin TEM-compatible brain slices) makes the proposal of such an "extreme neuroanatomy" project out of the question using today's slicing and imaging equipment. Regrettably, it is at this synapse-resolution that the brain operates. All the questions neuroscientists are asking relating to the computations performed in the brain can only be addressed at this level [1]. As an analogy, imagine you were given the task of making a street navigation atlas of Los Angeles. You could make such a street navigation atlas using a set of satellite images, but only if the satellite images had a resolution significantly better than the average width of the streets in LA. If you were given satellite images of LA having a resolution of 100 meters you would be able to locate downtown, major regional boundaries, and perhaps even the major freeways; however, such images would be totally useless in creating a street navigation atlas. Unfortunately, it is this situation in which we find ourselves in neuroscience today. All of our current atlases are several orders of magnitude lower resolution than the minimum needed to see the underlying, all-important neuronal wiring. Continuing the street navigation atlas analogy: Just as a thousand slips of paper containing driving direction to random points in LA cannot replace a well-organized, comprehensive LA street map, so too the thousands of "snapshot" in vivo tracing experiments (or confocal light-microscope 3D neural reconstructions and "needle in a haystack" single TEM images) may never be able to add up to the power of a single, comprehensive Synapse-Resolution Brain Connectivity Atlas. Sparse Synapse-Resolution Brain Connectivity Atlas ... Having defined what is meant by a Synapse-Resolution Brain Connectivity Atlas, we finally address what is implied by the "Sparse" prefix. From the above discussion it can be inferred that the imaging of an entire vertebrate brain at synapse-resolution is an enormous task. In fact, one can conservatively estimate such an undertaking to take thousands of years even for a mouse brain (see slice time vs. imaging time discussion). Thus a logical question to ask is "Must the entire brain be imaged to provide the usefulness of a connectivity atlas?" The answer is no. It is well understood that the circuitry of the vertebrate brain is highly redundant with each brain region containing thousands of copies of essentially the same circuit repeated over and over. These repeated circuits are not identical, having been modified by learning during an animal's lifetime; however, once the basic wiring diagrams of a few of these repeated circuits are determined (and the statistical variations due to learning are also determined by comparing several of the repeated circuits to each other) imaging the rest of the circuits would provide little more additional information (at least toward the types of questions today's neuroscientists are addressing). One need only image a tiny fraction of the whole brain (perhaps 1/10,000, see Single Brain Physical Slice Library Paper for discussion) in order to provide the necessary neuroanatomical foundation to support the vast majority of questions neuroscientists are addressing today. Unfortunately, this "tiny fraction" of the brain which must be imaged is distributed more or less evenly throughout the entirety of the brain's volume. Neuronal circuits local to each region of the brain, and the specific projections between those same particular local circuits, must be imaged in order to provide a complete connectivity atlas. It appears that only by using intelligently directed imaging (i.e. using initial probe images to guide further imaging sessions) can the volume of brain needing to be imaged be brought down by the several orders of magnitude theoretically possible given the brain's inherently redundant construction. An online, collaborative, neuroanatomical imaging experiment called a Single Brain Physical Slice Library is our proposed methodology to allow such intelligently-directed "sparse" imaging of the brain. The Single Brain Physical Slice Library infrastructure is designed to allow many remote neuroanatomical specialists the ability to decide how precious imaging resources are allocated in mapping the circuits of the brain. The result of this ongoing collaborative imaging experiment would be a Sparse Synapse-Resolution Brain Connectivity Atlas which would grow more complete every year of operation, and which would succeed in mapping the statistical regularities of all the regions, axonal pathways, and neuronal circuits of the brain after imaging only a tiny fraction of the brain's total volume.
[1] Of course neurotransmitters, ion channel types and distributions, and complex transduction pathways play a crucial role in all neural computation in the brain and can be considered to involve structures much smaller than even 10nm. However, these actors are best thought of as characteristics of classes of neurons, and can be best studied using in vitro brain slice pharmacological and physiological experiments. Experimental results of these in vitro experiments can be correlated with structure (for example, finding a particular structural characteristic of all synapses using a particular type of neurotransmitter) and thus used to interpret the 10nm resolution, structure-only images. It is an open question as to whether 10nm, Osmium-stained, synapse-resolution images implicitly contain all the information needed to infer the pharmacological class and genetic state of an imaged neuron.
|
Last Updated: 11/09/2003 |
