X-Message-Number: 3271
From: Brian Wowk <>
Date: Sat, 15 Oct 94 14:28:59 CDT
Subject: SCI.CRYONICS Brain scanning and uploading

        Thanks, Yvan, for your interesting message about brain 
uploading.  I would, however, like to correct some misconceptions 
about brain imaging.
 
        First, achieving sub-micron resolution with ionizing radiation 
(x-rays or gamma rays) is impossible without destroying the tissue in 
the process.  There are a number of ways to prove this.  One is to 
consider that resolving detail at a given voxel size requires *at 
least* one photon to be absorbed per voxel.  For example, a brain 
contains about (1 million)^3 = 10^18 voxels 0.1 microns wide.  To 
acquire an image with that resolution would require the absorption of 
at least 10^18 100keV x-ray photons, with a total energy of 10^23 eV 
or about 10,000 Joules.  For a 1kg brain, this is 10,000 rads of 
radiation-- a fatal dose!
 
        Another way to see the problem is to consider the number of 
photons that must be collected photographically (or by other means) to 
encode the 10^18 bits of information that represent a brain at this 
resolution.  Clearly, the number of photons collected must be on the 
order of 10^18, and each one of these photons will have had a some 
probability (near unity) of interacting with the brain on its way to 
the photograph plate.  Again we are forced to conclude that a 
prohibitively large radiation dose would be absorbed by the brain.
 
        The above considerations are general, and would apply *at a 
minimum* to any imaging modality (including holography) that uses x-
rays to image the brain.  In practice things are often worse.  For 
instance, in CT scanning the requirement for adequate counting 
statistics as resolution is increased escalates the dose even more 
rapidly than the above analysis would suggest.  In fact, there is a 
4th power relationship between dose and spatial resolution.  Imaging a 
brain at 0.1 micron resolution using CT would require 10^16 rads!!!!
 
        Things aren't much better with MRI either.  In your message 
you state that going from 1mm to 0.1mm resolution would require a 
1000-fold increase in scan time.  This is incorrect.  In MRI, noise is 
constant (depedent only on the temperature and sample size).  
Increasing the resolution by a factor of ten (decreasing voxel size by 
a factor of 1000) therefore decreases the signal-to-noise ratio (SNR) 
by a factor of 1000.  SNR can be recovered by averaging multiple scans 
over time, but only in proportion to the *square root* of the number 
of scans.  This gives rise to the famous 6th power relationship 
between scan time and resolution in MRI.  Increasing resolution by a 
factor of ten increases scan time not by 1000 times, but 1 million 
times!
 
          Now you know why brain backup and uploading are an 
intellectual fascination for cryonicists, but not a practical one.  
Talking about such things is like talking about star travel.  These 
things are important and they will eventually be achieved.  However 
they will not be achieved by means so crude as external imaging, and 
will not be achieved in our natural lifetime.
 
        Technologically, we are much, much closer to being able to 
indefinitely preserve our memories and identity in-situ than copy them 
elsewhere.
 
                                        --- Brian Wowk

Rate This Message: http://www.cryonet.org/cgi-bin/rate.cgi?msg=3271