X-Message-Number: 3318
From: Brian Wowk <>
Date: Fri, 21 Oct 94 12:39:13 CDT
Subject: CRYONICS Brain scanning, reply to Marvin Minsky

In reponse to my proposal for a MRI-based microscopic brain imaging system,
Marvin Minsky writes:
>Good.  This is very important!  Can you estimate the required
>geometry. For example, what resolution could we get with needle probes that
>have, say, 200 micron diameters that are inserted in a grip with 1 cm.
>spacing. I can imagine being able to do this in the not-too-far future
>with negligable brain damage.  That is, unless you do it a hundred or
>so times.  We'd have to insert the probes without causing serious
>microhemmorhages, of course -- but if the probes were active during
>insertion, they could be made to avoid all significant blood vessels.
        Each rf probe must consist of a loop of wire around an axis
that is perpendicular to the main magnetic field.  The loop can be
any size you want, but anything smaller than ~100 microns would
probably require a microscopic on-site preamp.  The field of view
of each coil would be equal to about two coil diameters.
        The spatial resolution would be determined by the strength
of the switched magnetic field gradients used to encode position
information as frequency shift during imaging.  A resolution limit
occurs because of the natural line width (inherent frequency spread)
of the tissue being imaged.  For instance, the most powerful
head gradient sets used in MRI today produce a gradient of about
0.0001 Tesla (1 Gauss) per millimeter.  At a main field strength
of 1 Tesla, the "chemical shift" frequency difference between protons
in fat vs. water is about 100Hz.  The result would be that lipid
parts of the image would be misregistered with the water parts of the
image by about 25 microns.  This misregistration artifact could be
dealt with by selectively exciting either fat or water, obtaining the 
two image sets seperately, and shifting them into registration later.
In such a scheme, the ultimate resolution limit would probably be
about 2 microns (determined here by the inherent bandwidth given by
the T2 relaxation time of brain tissue (~100 milliseconds).
        Improving on the above resolution would require stronger
gradients.  This is not possible for *external* gradient coils
because prohibitively large currents would be induced by dB/dt in
peripheral areas.  You would therefore have to put not just rf coils,
but high-current gradient coils inside the brain as well, which is not
something I care to speculate on.
        By the way, all this assumes that the brain is immobilized
*very* well during imaging.  The procedure would probably require
deep hypothermia and cerebroplegia-- i.e. complete shutdown of brain
blood circulation for prolonged period of time.  This can be done,
but it is very dangerous at present.
--- Brian Wowk

Rate This Message: http://www.cryonet.org/cgi-bin/rate.cgi?msg=3318