Synapse Resolution Whole-Brain Atlases

It is well-known that the highest resolution whole brain atlases are currently at BrainMaps.org, which has been compared to a Google Maps for the brain. However, these atlases are 0.46 microns per pixel, and are not sufficient to discern individual synapses, which require nanometer resolution. So in this post, I will consider the problems associated with constructing a synapse resolution (nanometer resolution) whole-brain atlas.

There seem to be two fundamental hurdles to constructing a synapse resolution whole-brain atlas: 1) image acquisition, and 2) digital technologies for working with the images and serving them over a network.

The first hurdle encompasses the time bottleneck and section preparation. If each section is 50 nm thick, then for a 10 mm mouse brain, 20,000 sections are needed, thus requiring some type of automation for section preparation. If we consider the time to scan a single 10mmx10mm section at 1MHz, it comes out to 46 days, which is unacceptable. Even with 20,000 TEMs (transmission electron microscopes) in parallel, one for each section, it will take 46 days for the complete scan. An alternative is offered by way of virtual microscopy solutions offered for light microscopy. One way would be to scan over the section, acquiring one column at a time instead of a patchwork of small images for montaging. Another alternative would be to construct a TEM with parallel scanning capabilities (having parallel magnetic lenses and electron beams), so that the entire section could be scanned at once, instead of scanning each little image patch in serial. This solution requires constructing a special type of TEM which implements certain features found in current day virtual microscopy systems for LM (light microscopy), and thus requires a team of hardware and software specialists to specially design, in addition to some physicists who are intimately acquianted with the physics behind TEM.

The second hurdle involves digital technologies, and the observation that even if a whole mouse brain was able to be acquired through TEM, that digital technologies currently would not be able to deal with that much data (8 x 10^17 pixels, or 2.4 10^18 megabytes (uncompressed)). A single section is 4 x 10^12 pixels, which comes out to 12 x 10^12 megabytes or 12,000 petabytes (uncompressed), which is still not feasible using today's digital technologies.

Let's consider a less ambitious proposal: TEM montaging of a 1mm x 2mm area at 2.5 nm resolution. TEMs typically acquire images in 2kx2k patches, which means that each patch is 5 microns x 5 microns. So for 2mmx1mm, it's 80,000 patches, and the montaged image size would be 800k x 400k, which is already a problem since there are file format size limitations on common formats like TIFF and JPG, and so to acquire such a large image would necessitate using a non-standard file format, which makes the issue of making the images web accessible more problematic. The largest images, say at BrainMaps.org, are 120k x 100k, which works out to 3 GB as a JPG-compressed TIFF file (or 30 GB uncompressed), and which is already near the limit for the TIFF file format (which is 4 GB), which means that images much exceeding 120k x 100k are already going to present a problem.

In conclusion, for purposes of obtaining information about whole-brain connectivity, a nanometer-resolution whole-brain scan is required, and current-day tracer experiments are suboptimal and will always leave room for ambiguities that can only be resolved by completely mapping every synapse and axon in the brain. However, constructing a synapse resolution (or nanometer resolution) whole-brain atlas for even a mouse brain is so formidable as to be seemingly beyond today's technological capabilities. Maybe in 10-20 years.

Paul Allen Brain Atlas Misconceptions

I had noticed last Tuesday a blip on the visitor activity site statistics for this blog, and when I looked into it further, saw that increasing numbers of people were coming to this blog by searching for the Allen Brain Atlas. Apparently, this visitor activity 'blip' corresponded to a publicity campaign launched by the Paul Allen marketing department on that same day to publicize that all the genes in the mouse brain had been mapped.

I have posted some things that were critical of the Allen Brain Project, but not unrightly so. I want to see the project succeed and not merely create illusions and spread disinformation through the media. With this in mind, I would like to correct some the media hype and falsehoods about the Allen Brain Project that have been widely circulated.


Common Allen Brain Atlas Misconceptions

1) The Allen Brain Atlas will contain over 1 PetaByte of data.

False. The orders-of-magnitude calculation was done by multiplying 20,000 genes by a trillion neurons, but this is a gross overestimate. A more realistic computation involves multiplying the number of datasets they have, which is around 20,000, with the average size of each dataset. The average size of each dataset is about 10 slices, times the size per slice. The size per slice is about 10,000 pixels wide, which works out to 100 megapixels per slice. Without image compression, each megapixel is 3 megabytes (one byte for each color channel), which means that each slice is 300 megabytes, uncompressed.

Thus, a more realistic calculation of the size of the Allen Brain Atlas is
(20,000 datasets)*(10 slices per dataset)*(300 megabytes per slice) = 60 TeraBytes.

So, the real size of the Allen Brain Atlas is around 60 TeraBytes, which is a far cry from a PetaByte.


2) Since mice and humans share more than 90 percent of genes, the Allen Brain Atlas has enormous potential for understanding human neurological diseases and disorders.

False. We share over 70% with insects and over 50% with plants, so according to the logic of the Allen Brain Atlas people, dissecting the genetic maps of oranges can help us fight heart diseases and schizophrenia.


3) The Allen Brain Atlas will provide the most detailed map of the most complex organ.

False. http://BrainMaps.org provides the highest resolution whole brain maps, and not just for mice, but for primates and other species. The resolution of BrainMaps.org data is over twice as good as that of the Allen Brain Atlas.


4) The Allen Brain Atlas has already led to several significant new findings about the brain.

False. There are absolutely no peer-reviewed publications over any significant new findings from the Allen Brain Atlas. I do believe that significant findings can be made, but there is nothing published about it in peer-reviewed articles as yet. (Update!: in Jan 2007 they did finally publish one article; unfortunately, it contained nothing new and what they were presenting as "new" was in fact old work that had been published by one of the authors, Lein ES, well before the inception of the Allen Brain Atlas.)


5) The Allen Brain Atlas provides a complete genetic map of the mouse brain.

False. It says nothing about silent DNA or junk DNA, not to mention splice variants. Nor does it say anything about genes involved in development or disease, nor about gene expression variance or genetic interactions. Plus, the fact that a large percentage of their probes failed (which means they don't have data for genes they claim they have data for), and in many cases, the data is either wrong, or it's very poor quality.


6) The Allen Brain Atlas gets more than 12 million "hits" a month.

False. As of Feb 24, 2007, the Allen Brain Atlas ranks in at an abysmal #1,173,023 according to Alexa, which means that they bring in no more than a few hundred visitors per day. In other words, the Allen Brain Atlas is a relatively unpopular site.


7) The Allen Brain Atlas utilized factory-like efficiency.

False. Over a period of just three years, over $42 million was spent on the Allen Brain Atlas, which is outrageous (thankfully, taxpayers were not footing the bill). There were 40-50 people employed for the project, which means that probably $4-5 million per year went directly into their pockets for salary. In all seriousness, had the Allen Brain Atlas utilized factory-like efficiency, then there would have been around 10 people employed and the total cost would have been kept between $3-9 million.


8) The Allen Brain Atlas is "Epoch-Making".

False. The Allen Brain Atlas is one of many projects that aim to better map the brain. As such, trying to present the Allen Brain Atlas as the only large-scale "epoch-making" brain mapping project is ludicrous and reflects the self-centeredness and disengagement from reality of the project managers making such claims. I have posted previously about how many of the original prominent people involved with the project deserted it due to mismanagement, disenchantment, and inner power struggles. The problem is that the ideal that originally guided the project three years ago is being substituted for the reality of what the project actually is today. And what it actually is today, while useful, is not all that unique, and certainly not epoch-making by any stretch of the imagination. The disinformation being pumped into the media is something that should raise concerns; my own personal thoughts are that this media disinformation campaign is the result of a lot of money being thrown around. Maybe that's where a lot of the $42 million went that should have been going towards the project, quality control, and better user interfaces.

Google Earth for the Brain

If BrainMaps.org is like Google Maps for the Brain, StackVis is Google Earth for the Brain. Welcome to StackVis, a 3D viewer of neuroanatomical sections.








Figure Caption. The development of additional desktop application tools for interacting with brainmaps.org image and database data includes the one shown here, StackVis, which is a 3D viewer of neuroanatomical virtual slide image stacks that is integrated with high resolution viewing of and interaction with individual sections comprising the image stack. (A) horizontal image stack of nissls of the macaque brain viewed from below. (B) same image stack as in (A) but from a different perspective, with increased inter-section spacing, and with areal and nuclear labels. (C) coronal image stack of nissls of the mouse brain. (D) a section from (C) viewed at higher resolution. (E) coronal image stack of acetylcholinesterase reacted sections of the mouse brain. (F) sagittal image stack of the mouse brain reacted for biotinylated dextran amine (BDA) following an injection in frontal cortex. Note that labeled fibers can be followed within the image stack due to section transparency and that individually labeled subcortical structures can be discerned, allowing for an assessment of labeled fiber pathways and areas within a 3D framework. In addition, StackVis features automated section registration and edge detection capabilities.

Figure Caption. Viewing the Visible Male using StackVis. The sections are axial, and are arranged in this figure so that the brain is located at the top and the legs are located at the bottom of the image stack.

Virtual Microscopy a Disruptive Technology?

Virtual microscopy is a method of posting microscope images on, and transmitting them over, computer networks. This allows independent viewing of images by large numbers of people in diverse locations.

Prior to recent advances in virtual microscopy, slides were commonly digitized by various forms of film scanner and image resolutions rarely exceeded 5000 dpi. Nowadays, it is possible to achieve more than 100,000 dpi and thus resolutions approaching that visible under the optical microscope. This increase in scanning resolution comes at a price; whereas a typical flatbed or film scanner ranges in cost from $200 to $600, a 100,000 dpi slide scanner will range from $80,000 to $200,000.

Virtual microscopy has been characterized as potentially a disruptive technology. A disruptive technology is a new technological innovation, product, or service that eventually overturns the existing dominant technology in the market, which in this case would be real (i.e., conventional) microscopy. Our experience with virtual microscopy suggests that it is unlikely to replace real microscopy any time soon, but for the time being, it nicely complements and extends the capabilities of real microscopes. Specifically, we find a three-fold extension of virtual microscopy over real microscopy in the following areas: 1) data-sharing and remote access, 2) data-management and annotation, and 3) data-mining. Data management and data-mining of virtual (digitized) slides are capabilities that cannot be directly applied to real slides. In addition, the online distribution and sharing of virtual slides with anyone with an internet connection ensures the rapid dissemination of neuroanatomical data that otherwise would not be possible. While largely emphasizing the pros of virtual slides in this article, it is worthwhile to point out the cons. Namely, it is not possible to change focus in a virtual slide as it is in a real slide. Normally, this is not a problem since virtual slides tend to be completely in focus. However, the inability to change the plane of focus in a virtual slide rules out their use in unbiased stereological estimation methods that require optical dissectors. Nonetheless, biased sterelogical estimation methods, or unbiased methods not using optical dissectors, are still possible. Another drawback is that the resolution of the virtual slide is limited to the optical lens used in the scanner. For example, if we generate a virtual slide at 20x and subsequently want to examine part of the slide at 40x, then it is necessary to rescan the entire slide using the higher objective, which in some cases, is not possible due to file size restrictions or hardware issues. Finally, at the time of writing, virtual microscopy does not deal well with fluorescence, and is only recommended for light microscopy.

Virtual microscopy-based digital brain atlases are superior to conventional print atlases in five respects: 1) resolution, 2) annotation, 3) interaction, 4) data integration, and 5) data-mining. The resolution of conventional print brain atlases typically does not exceed 7200 dpi, whereas virtual microscopy-based digital brain atlases attain 100,000 dpi and offer the ability to zoom in and out. Annotation can be more complete in virtual microscopy-based digital brain atlases, with options to display some types of annotations and make the rest invisible. Greater interactivity means that the user can zoom in/out and pan through brain image data, which is not possible in print-based atlases. Data-integration capabilities, including the integration of connectivity and gene expression data, are superior for the virtual microscopy-based digital brain atlases. And finally, the ability to data-mine virtual microscopy-based digital brain atlases is a feature not available for print-based atlases. While we do not foresee virtual microscopy-based digital brain atlases completely replacing conventional print-based brain atlases, we expect that they will be progressively more commonly used in place of print-based atlases.

What I'd like to know is, how likely is virtual microscopy to overtake regular microscopy and print atlases in the near future? In my opinion, conventional print brain atlases are a ripoff, often costing over $200 each. The idea of having this information freely available online is very tantalizing and welcome.

In other news, Society for Neuroscience 2006 is fast approaching! Hope to see you all in Atlanta this Oct 14-18.

Collaborative Digital Brain Mapping Comes of Age

Google Maps and related geomapping services provide high-resolution satellite maps to anyone with an internet connection and have set the standard for online digital mapping. We are now beginning to witness similar digital mapping technologies spilling over into other non-related fields, one of the more interesting of which is neuroscience and the collaborative digital mapping of the brain.

Launched less than a year ago, BrainMaps.org has rapidly developed to lead the field in digital brain mapping technologies. With several terabytes of ultra high-resolution brain image data, consisting of several dozen mouse, monkey, and human brains, its online brain image database is the largest and most diverse currently available. This massive image data is integrated with structural information regarding spatial locations of different brain areas and markers, and the relations between them. And in the collaborative spirit, online users are free to add their own labels and annotations, and to place landmarks throughout the digital brains they explore. Users may even share their images, landmarks, and other annotations with other users in the BrainMaps forum, which in many ways parallels the Google Maps Community, but on a smaller scale.

The U.S.-sponsored 'Decade of the Brain' has come and gone; it officially ended in the year 2000. It would take another five years before BrainMaps.org came onto the scene, and in a way, it encapulates what the Decade of the Brain should have been about: Collaborative digital brain mapping and a resource available to everyone with an internet connection.