From: Eugen Leitl Newsgroups: sci.cryonics Subject: neuropatients&uploaders Date: Tue, 6 Jun 1995 19:25:59 +0200 Organization: Leibniz-Rechenzentrum, Muenchen (Germany) Distribution: world NNTP-Posting-Host: sun4.lrz-muenchen.de Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII X-Sender: ui22204@sun4 To: cryonet 0. Motivation Uploading pursues distinctly different goals than the cryonics mainstream. Hence, it will also require somewhat different procedures. This is an informal outline of some of them. 1. Death and Euthanasia This is primary a legal and medical ethics question. If we extrapolate current relaxation trends ("anything goes") into the future, there seems to be some reason for cautious optimism. Adopting citizenship of legally pioneering states shortly before imminent death may appear as a sensible idea. Nonlegal practices should be avoided at any costs: demand for trust and transparency of a cryonics corp will require easy inspection of financial and storage conditions, which is impossible for black clinics. Since successful legal action can cause instant demise of affected cryonic corp and its entire patient bank, this requires very careful legal navigation. A decephalized body can be released almost immediately to relatives for rites/burial with relatively small, easily camouflageable superficial damage, a possibly important point for those wishing to conform with relative's expectations. In future euthanasia will be certainly allowed for persons laboring under an intreatable illness in most countries of western hemisphere. Even suicide may become legally allowed with (some) time. Of course this will make life insurance financing impossible, thus requiring alternative financing schemes. Induced cardiac arrest (e.g. IV KCl-triggered) of (precooled) anaesthesized (freon- or dinitrogenoxide-type gas inhalation anaesthesy) patients seems to be a good model for painless euthanasia. 2. Uploader/Neuropatients Reduced tissue bulk size to be processed as required for neuropatients has very definite advantages: The total bulk to process, transport and store is much smaller. Thermal energy content of tissue volume to process is smaller. Tissue is monotyped/isotropic. Speed of cryoagent diffusion is excellent, less agent is necessary. The necessary cryoprocessing machinery can be very compact, cheap and easy to transport (a major cost factor especially for oversea actions). Cryoteam size can dwindle to 2-3 persons. Neuropatient transport is trivial. Legal ("tissue sample" vs "corpse") as well as freight volume pose basically no problems. Dry ice exhalation or lN_2 boiloff may cause difficulties during air transport (cabin air contamination with nonbreathables), though. A Peltier cool box may be better here. Brain's surface/volume ratio is much better, allows much higher initial (process bottleneck) cooling rates. The brain has excellent capillary infrastructure, allowing thermic descent and perfusion in one go. Reduced storage volume is another big plus. At the same dewar size orders of magnitude more patients can be stored, allowing much lower per capita ;) cooling medium cost. 3. Decapitation and Cryoperfusion Programme Warm ischemia period should be minimized. But, since we are going for information death time window here, constraints are not so very. After instant decapitation (hydraulic/electric guillotine?), cryoperfusion outlets should be immediately plugged into the main neck arteria(s). Diameter-adjustable no-fuss custom clamp outlets will be required. All blood must be flushed out instantly. Anticoagulant and initial flush solution composition/temperature must be optimized. Plastic wrapped (multiple layers) head can now be submerged into an external ice/water bath to assist internal cooling. A cheap peristaltic pump (redundant, of course) may suffice here. After all blood has been flushed out, closed circuit flux can be established. The oxygenated perfusion fluid can be the standard one as for organ preservation, though alternative receptures should be investigated. A steep temperature descent programme, using RT/0 deg C gradient mix machine is then initiated. After arriving at 0 deg C, the cerebrum is removed from the skull and reconnected to perfusion jacks. After 5-10 min, 0 deg C temperature should be achieved. Now cerebrum is suspended in transit flow container for oxygenated cryofluid (plus glucose) and a cryofluid gradient programme is begun. Through low-toxic, pure glycerol diffusion rate is suboptimal. Dimethylsulfoxide (DMSO, Me_2SO_2, m.p. 18.4 deg C) is a great transmembrane shuttle and breaks water structure even better than propantriol at low tissue toxicity. As lots of phospholipid membranes have to be permeated, DMSO is orders of magnitude better agent than propantriol. A mix of both should be probably optimal, I think. After thorough permeation has been achieved, (vacuum) deoxygenated (helium/argon flushed) fluid, now without other additives is used for perfusion. Since radical chain reactions work best at deep temperature, very low oxygene residual content must be achieved. Glass matrix will prevent the worst, yet... Now temperature descent gradient to liquid N_2 temperature is initiated. Cracking should be minimalized, yet is not very critical, since we are going to cut the tissue up in the end all the same. Entire procedure and personal data are stored in the corp database, printouts on acid free paper go into archive. The same data, recorded on a sturdy medium (photolitho metal foil or photo-glass microfiche) in human readable form should be stored with each patient. Additionally, digitized detailed picture material describing body geometry (high resolution color snapshots from different sizes, microphotography to document texture/colouring etc. en detail, skin/hair/other tissue specimens, voice samples, etc.), MRI/squid imaging data, and personal data should also go into the vault. Redundant storage medium should be a WORM, with periodical checks/rewrites to counteract medium deterioration. As much context as possible should be preserved, to reduce future sensory overload shock after reanimation. 4. Tissue Storage Each cerebrum should be sealed (in a vitreous cryofluid block?) in oxygen-free microenvironment together with basic documentation in a carrier with both machine-readable (e.g. bar code (rime is transparent for IR) or induction loop sender (won't work, if too cold)) and human-readable label. Carriers should fit into columnar metacarriers accessable from dewar top. Each dewar should have a machine/human readable list linking position index with patient ID to facilitate retrieval. Dewar volume should be big to enhance volume/surface ratio. Though spheres are best, cylindrical shape does greatly facilitate de/loading. Since for good dewars black body radiation is the major energy leak, external cooling of outer dewar layers (e.g. to -20..-70 deg C, the lower the better) using standard (molbio lab fridges) refrigerator technology could significantly reduce nitrogen boiloff. Sufficient -70 deg C external refrigerator capacities should exist for emergency. Though industrial lN_2 is much cheaper, a local diesel-driven air rectifier should be considered as failsafe cryomedium backup source. Photovoltaics and/or geothermal gradients are better than diesel, though. Conservative estimates seem to indicate a monocrystalline Si solar panel to usable for at least 50 years. Particularly, air/subterraneous water stream temperature delta could be used for direct cooling (Carnot cycle machine). Though having bad yields, Peltier element arrays should be discussed as 1stage cooling/energy source. 5. Tissue Fragmentation Vitrified cerebrum bulk must be cut (diamond or glass knife microtome) into cubic tissue blocks, edge < 1mm. Edge cut artefacts have to be removed with DSP methods. Both block x,y,z index and orientation must be recorded, e.g. on the block carrier. Such tiny specimens can be temporarily thawed to be subject to immunostaining or other contrast-enhancing procedures. Each block can be scanned independantly concurrently. 6. Scan/Processing 6.1 Nondestructive Scan Yet-to-come xRay fresnel/holographic diffraction lattice optics utilizing monochrome synchrotrone radiation might do. But: the contrast ratio is very poor, immunostaining with metal-labeled antibodies may be needed. Energy deposit by xRay absorbance will cause cumulative rad damage, virtually cooking the sample. Gamma cannot be focused at all, apart from requiring nuke-pumped lasers for sufficient intensities. Since brain is no crystal, there is no diffraction picture to analyze. Tomography might be possible, yet resolution is not so very great. Neutron beam tomography is another alternative. Suffers from the same problems. Transmission EM has sufficient resolution, but requires very thin samples with very elaborate preparation. Rad damage here also. This is about all. There are not any nondestructive scan methods I can think of. Any ideas? 6.2 Destructive Scan My long-time favourite has been UV-abrasion of immunostained vitrified tissue coupled with sensitive mass spectrometry. A weak pulsed excimer laser output, channeled through a quartz (or something with wider UV window) optofiber with 50-100 nm tip (no optics, since we cannot focus spot beyond wavelength) shining upon a point of cooled vitrified probe in vacuum will instantly dissotiate bonds in a layer few molecules thick, causing ionized debris expand isotropically. A sensitive MS can suck off this cloud, analyzing debris. Nanoantibodes (essentially, a tiny antibody fragment having only one antigen-bonding site) can be easily (nonradiocative) isotope labeled, offering rich CHNOS (c12/c13, h1/h2, etc.) isotope fingerprints. Engineered microorganisms, grown in isotope labeled media are simple (but not cheap) sources of such antibodes. Alternatively, abrasion could also be done with an electrone beam writer. Vertical resolution goes down, though. Resolution is not too great, but a wealth of detail can be read since isotope labeling allows plethora of label flags. Abrasive atomic force microscope (AFM) scan is much better It needs no vacuum nor any sample preparation, reads and abrades with the same medium (needle) at up to video data stream rates and has excellent resolution (single atomes/molecules). But it produces only the shape information ("form follows function"). The device is very cheap (<$500-$1000), if mass produced. If scanned tissue is not too cold (ice will scratch metal, if sufficiently cooled), needle tip will serve almost indefinitely. 7. Storage Let us assume 10 G neurons with the average synapse branching factor of 10 k (though in single cases convergence factors of up to 100 k are known). Hence a 34 bit neuron ID is sufficient. Each neuron body state dynamic range (firing probability or mean firing frequency within time window) will be represented by an integer of 0..255 (8 bit), each synapse weight 8 bit and 8 bit for signal delay. Hence, each neuron will need 8(+8)+(34+8+8)*10 000 = 0.5 MBits for an adequate representation. In toto 10 G * 0.5 Mbits=5*10^15 or 5 PBits (peta bits). Assuming off-shelf 10 GBytes (80 Gbits) DAT cartridge, we'd need roughly 64k cartridges. Taking soon-to-arrive 1 TByte (8 TBit) cartidges, about 600 of these. Notice that this is the maximum amount. This is uncompressed data: utilizing representation redundancies (e.g. relative instead of absolute addressing) at least one order of magnitude compression ratio can be achieved. Moreover, 6 bits instead of 8 is probably the dynamic range. Delay is probably unnecessary, since message packet routing artefacts introduce inherent delays, proportional to original physical distance. So 80 1TByte- or 8000 standard DAT cartridges is probably the number in question. Notice that was lossless compression. Higher-order codings, sufficiently elaborate to accurately mimick state-space strange attractor kinetics could probably reduce the amount to a 8-80 DAT cartridges. But this is pure blueskyeing. I have no data on cerebral system collapsibility whatsoever. 9. Upload A. Modus operandi