Background Material for Chapter 6
POTENTIAL IMPLICATIONS OF BRAIN IMAGING RESEARCH
Behaviors requiring complex integration of several cognitive and motor tasks are core in today’s technological world. As our vision and scientific endeavors grow toward Bunyanesque undertakings, we push the envelope of the cumulative stress normal humans can be expected to absorb. Current soldier-system monitoring includes occasional measuring of physiological parameters as well as recording of performance observables following training and familiarization exercises. These techniques are little advanced from clipboard, stopwatch, and fill-in-the-blank subjective rating scales. Although effective at diagnosing general health and skills in combat maneuvering, hand-to-hand combat, and precision fire, these modalities are incapable of providing useful information to prevent the onset of neuropsychiatric conditions before they result in degradation of performance and thus increase the risk of human error.
Recent evidence from operations Desert Shield and Desert Storm and current Iraq operations indicates unequivocally that such concerns are more important now than in past conflicts. Their being accorded greater importance may be due to greater vigilance being paid to field performance under close quarters with better real-time observation, or to a new awareness of the importance of the antecedents to post-traumatic neurosis as a real psychodynamic to an accurate diagnosis, or both. It may also be a condition of the changes in the way we now ask individuals to perform under stress, with increased individualization in both decision making and performance while directly in harm’s way. The current enemy and that of the next three decades have an advantage in serving under a sociopolitical-religious mandate unacceptable and unfamiliar in the West.
Finding new ways to elicit, observe, and record behaviors to ensure the continuity of health over long periods of confinement and isolation is a key research area. In addition, decision-under-stress reactions are native to each individual, adding complexity to human factors analysis during traditional training schemes as well as to monitoring of mental acuity. Extra design considerations are also required to fold in the cumulative effects of very long (1 to 2 years) mission durations in harsh climates where stress takes a continuous toll, and of less-than-war situations, in which death nonetheless awaits literally
at every corner, every day. Finally, careful and appropriate attention is needed to psychiatric liability, informed by past experiences, which can at best confound and at worst imperil such missions (Genik et al., 2005, in press).
Current and future tools and techniques in brain imaging promise to provide essential additional information to monitor the effectiveness of training as well as mental status. The recent advent of functional neuroimaging, discussed below, has sparked world interest in the four major noninvasive brain-monitoring modalities: electroencephalography (EEG), magnetoencephalography (MEG), functional magnetic resonance imaging (fMRI), and near-infrared spectroscopy (NIRS). In this appendix the committee explores the current technologies and how in the future they may be engineered and employed for use for monitoring in simulated (training) and actual pre-mission qualification, especially of the young officer cadre entering battle under nontraditional and high-threat conditions. The methodologies are amenable to application in the assessment of mental status and training efficiency, under flexible conditions and in virtual-reality testing and training scenarios.
ENABLING TECHNOLOGIES FOR A NEW BLUE FORCE CAPABILITY
Here the committee postulates a new BLUE force capability that enables training and networked communication, as well as interview/interrogation. It envisions a portable modular sensor suite that is:
Expandable and flexible, and
Compatible with current systems.
Current methods for human-human tactical communication must be informed by new cognitive research. In the next decade, it may be possible (with minimal military-specific operational integration of the following technologies) to determine with specificity nonverbal communications between and among soldier systems. The same research will inform a variation of this capability to replace entirely the current hostile, putatively unethical, and near-valueless physical coercion techniques used to determine truth and deception in captured RED forces, both enemy soldiers and nonmilitary combatants.
Enabling technologies include emergent functional brain neuroimaging methodologies together with data fusion software applied to multisensor, multispectral neurophysiological signals.
Potential applications include expanded interpersonal communications via BLUE soldier-system wearable body and helmet mounts; multi-lead covert and overt polygraphy applications and psychological interview techniques; and detection of intentions, deception, and truth in captured RED combatants.
Subsequent sections describe in greater detail the emerging technologies that may enable such capabilities.
Functional Neuroimaging Technologies and Soldier Systems
Human brains are constantly occupied with tasks that include unconscious activities such as regulation of breathing and circulation, and highly cognitive functions such as reading and interpreting technical reports. The human brain does not take well to a complete reboot to a known state, and therefore the measured state of any individual brain is a function of its initial state (genetic makeup) and the entirety of internal chemical and external biosensory stimulation integrated from birth to the present. To overcome the extensiveness of variation, one can take a picture of the brain in a baseline state and immediately afterward apply some external stimulus or have the subject perform a simple task and take
a second picture in the activated state. The difference image will reveal what parts of the brain were activated to confront the stimulus or perform the task. This is the principle behind functional neuroimaging. In practice, baseline and activation states are repeated and recorded several times to average over inherent noise and prove repeatability. Details about how many repetitions are required depend on the imaging method, or modality used.
Electroencephalography is a compendium of measured voltage differences at the scalp obtained from between 10 and 300 probe electrodes. Neuronal activity in gray matter or white matter axonal transmission of signals, the primary current, induces cascading volume currents in surrounding tissue and ultimately surface currents on the skin. The electrical diffusion is very fast, and for all practical purposes the signals that appear at the scalp are simultaneous with the brain activity. The best spatial resolution is obtained for cortical primary currents. Deeper subcortical primaries are detectable as well with a corresponding decrease in spatial resolution. When used in a functional experiment, the measured voltage temporal changes related to a task or stimulus are called event-related potentials (ERPs).
Involved in construction of an EEG is the electrode pack, which can be a personalized flexible cap, a set of signal amplifiers, a multichannel few-hundred-hertz digital sampler, and an interface multiplexer, such as the next-generation serial bus, for presentation to a standard interpretation display, recording device, or general-purpose onboard computer. When designed to be lightweight and compact using even current technology, a 64-channel EEG can be easily incorporated into a tanker-training module, or into the actual tank, APC, Bradley Fighting Vehicle, BCW enclosed FUCH’s scout vehicle, helicopter trainer or actual cabin, and more. Indeed, an EEG system can be incorporated into advanced extravehicular mobility units (EMUs) and integrated with the standard physiological telemetry for nominal additional weight in the new soldier-system dismounted uniform.
Magnetoencephalography measures near-scalp sub-picotesla magnetic field variations due to the time-dependent primary and volume currents using a cryogenic helmet-mounted array of hundreds of superconducting quantum interference devices (SQUIDs). MEG is similar in temporal resolution to EEG. The sensor array is designed to detect currents parallel to the scalp, and MEG excels in distinguishing these from currents perpendicular to the scalp (predominantly cortical surface currents in gray matter convex folds, or gyral currents).
The additional weight of precision electronics and control systems necessary for MEG is hard to justify if the only benefit is real-time separation of gyral and cortical surface currents, since individual astronaut EEG meshes can be calibrated with a preflight MEG such that the EEG can determine the area of activation with acceptable accuracy. However, the higher accuracy is justifiable if some of the SQUID measurement, control, and analysis systems can serve as dual-purpose apparatus for other experiments.
The brain activity measured by EEG and MEG is the result of ionic motion within the central nervous system. Ionic production, release, and movement consume energy. This energy is supplied by the conversion of blood-borne oxyhemoglobin to deoxyhemoglobin. Through a complicated chain reaction called the blood-oxygen-level-dependent (BOLD) effect, brain activity results in a net increase in the relative concentration of oxyhemoglobin in the local venous structure of the circulatory system. Oxy- and deoxyhemoglobin have different magnetic susceptibilities and different colors, giving rise to two methods to measure the local hemodynamic response to brain activity.
Magnetic resonance imaging (MRI) works by a simple excitation and relaxation of spin states. When molecules containing hydrogen are placed in a strong static magnetic field, a small but detectable number of hydrogen protons align their spins along the direction of the external field. An applied radio-frequency (RF) pulse near the proton resonant frequency knocks the spins perpendicular to the field, and the relaxation back to ground state releases RF energy in a pattern that can be reconstructed to show both
the composition and the distribution of blood, soft tissue, or any other hydrogen-rich material. A series of fast scans calibrated to optimize detection of the BOLD signal will show the dynamics of brain function under the specific internal or applied conditions; this is known as fMRI, and the major advantages of this modality are three-dimensional spatial resolution and complete skull penetration, making it the only modality to unambiguously detect limbic activations important for determining certain neuropsychological states. In addition to measuring the BOLD signal at higher precision, future hardware advances will allow fine temporal monitoring of neurochemical movement and reactions using ultrafast spectroscopic analysis techniques in fMRI.
Complete commercial clinical MRI scanners weigh as much as an entire communications satellite and are inappropriate to bolt to a military vehicle or helicopter bulkhead. What is promising, though, is that many of the bulky or heavy components of commercial scanners are not necessary in the embodiments that will be deployed: lightweight skullcap monitors with neural-network software based on the ground-heavy MRI scanning systems, and the MEG near-field applied to near-infrared spectroscopy and ported into existing low-power supplies, will be data-, not weight-, dependent. Additionally, whole-body toroid and multiscan high-resolution options are not required, but rather just a system big enough to fit a head with a single detection coil of moderate size because high-resolution scans can be collected pre-mission and digitally carried along. Functional MRI scans are themselves moderate-resolution, and any clinical diagnosis for traumatic injury to a scannable body part can be accomplished with a combination of high- and moderate-resolution images. Finally, the cryogenic components and superconducting magnet energizer or DC electromagnet power supply can be parasitized off other necessary onboard systems, vacuum pumps can be replaced with a good valve, and several ancillary failsafes that make the scanner clinician-proof can be eliminated. All told, an appropriate MRI scanner could be put together in a couple of hundred additional pounds.
Near-infrared spectroscopy, the technology behind the nascent real-time finger-clamp pulse oximeter, has developed into a commercially available 48-channel BOLD-signal-monitoring head array mounted in a flexible cap. The system currently has moderate spatial resolution for cortical activations. Future systems promise better resolution and deeper activation detection. Figure E-1 shows the neuroimaging modalities explained above.
The NIRS modality can be built using available technology and dual-task post-processing equipment. The sensory array is simply pairs of light-emitting diodes (LEDs) and photodiodes either directly mounted in a cap or remotely working through optical fibers. Indeed, the bulk and weight of the detection array make NIRS, like EEG, an appropriate consideration for integration into advanced EMUs as well as vehicle-mounted systems.
Further details of technical methodologies are available in an extended treatise (Genik et al., in press1) or complete reviews (Hamalainen et al., 1993; Jezzard et al., 2003). In summary, the above modalities are best when used in combination, and not necessarily simultaneously, as their neuroimaging strengths complement rather than duplicate one another. EEG excels at monitoring gross mental load in steady-state conditions, though it is also adequate for determining activity in pre-mapped cortical structures. MEG can discriminate between gyral and surface currents with moderate spatial resolution, and this can be vital information if a specific brain network structure is located in either of these places. MRI is the only noninvasive modality that can specifically localize deep-brain neural systems responsible for emotional processing, essential information for determining neuropsychiatric health. Finally,
NIRS provides a relatively inexpensive and robust monitor for cortical activation, a metric important for determining cognitive response efficiency at any instantaneous gross mental activity level.
Neuroimaging and Applications to Crewmember Health and Training
Beyond a discussion of biomedical engineering considerations of brain-imaging modalities, it is vital to determine whether information gathered offers important insight into the health continuum of crewmembers in military vehicles or of the dismounted, but in the future highly instrumented, soldier. The bandwidth will exist, the channels will exist, and the readouts will be distributed to the company level. The first comprehension of the potential for neuroimaging technologies in the context of long-duration spaceflight, as one example, was discussed in a 2001 NASA Jet Propulsion Laboratory workshop.2 The findings are identical to those expected for the current soldier in isolation, perceived isola-
tion, and/or continual stress. Moreover, the Bioastronautics Critical Path Roadmap identifies three mission confounders of particular relevance to the current Iraq operational scenario, for example: psychological disruption (e.g., circadian and sleep disorders), neuropsychiatric dysfunction (affective disorders), and nonspecific stress. All are amenable to current emergent neuroimaging modalities and can be studied in real time under fairly realistic conditions. Monitoring behavioral health of the crew can provide important input regarding mission decisions. Furthermore, pre-mission training programs will benefit from the additional insights into crew performance provided by neuroimaging.
Previous soldier long-duration mission training and EEG experiments have concentrated on narrow hypotheses, and the modality has not been adopted as standard procedure. Normal sleep is easily monitored with EEG, and deviations can raise alerts to possible developing situations. Waking EEG measurements show level of alertness and thus can help to determine risk levels for starting or continuing a mission-critical activity. Additionally, sleeping EEG traces have been associated with predictive outcomes for treatment of depression (Hatzinger et al., 2004), waking EEG traces have been associated with post-traumatic stress disorder (Chae et al., 2004), and functional EEG and changes in ERP have been used to monitor attention level in complex cognitive tasks (Ramos-Loyo et al., 2004). The next generation of studies will likely yield insight into many underlying affective disease mechanisms and metrics of behavior such as vigilance. Generation-after-next work will yield predictive algorithms from EEG sleep traces to head off circadian- and sleep-related disorders before they affect waking behavior (failure of human performance because of poor psychosocial adaptation or neurobehavioral dysfunction).
MEG has advantages over EEG for specific monitoring tasks (Barkley, 2004). Work completed in collaboration with Young et al. (2005)3 concentrates on the basics of piloting motorized vehicles while facing cognitive challenges, where MEG was used to locate functional activity related to event detection. Current work focuses on increasing the cognitive load and seeks to determine thresholds for overload, the point at which event detection breaks down and a driver or pilot can make a fatal error (failure of human performance because of system interface problems and ineffective workload). The next 5 years promise advances in MEG methodology as a complement to both EEG and fMRI, although it will likely be useful only for ground-based preflight training and screening of crewmembers. Generation-after-next technology will be required to start in-flight monitoring using this modality in a regular behavioral health regimen.
Advances in understanding of brain networks involved in piloting motor vehicles and high-performance aircraft based on use of the maturing technology of BOLD fMRI will drive functional neuroimaging understanding of these activities in the next 5 years. Previous studies have focused on narrow aspects of navigation (Uchiyama et al., 2003), attempts to localize impairment of navigational abilities (Calhoun et al., 2004), and the beginnings of next-generation understanding of event detection in simulated driving tasks (Graydon et al., 2004; Young et al., 2005). Current work is focusing on increasing the level of distraction and monitoring the effect on detection of the primary event. It is well known that even heavily sedated primates show activation in the visual cortex as a result of optical stimulation; it is also well known that humans can be looking directly at events and not react to them, a behavior known as cognitive blindness, or colloquially as the “look-but-do-not-see” phenomenon. What is not known at all is what disruption in which brain network causes cognitive blindness, as well as whether the risk of this phenomenon can be consistently modulated (Reilly, 2003). Of special relevance
here is the identification friend-or-foe problem set: friendly-fire deaths occur regularly in unmanned combat air vehicle (UCAV) and umanned aerial vehicle (UAV) maneuvers, helicopter tactical operations, sniper operations, and mounted Bradley and Abrams operations. Technology to understand and mitigate the “brain internal” disconnect in looking but not seeing, or comprehending, or processing may be more important statistically than a next-generation set of sensors that must be highly networked and computer-centric on the distributed battlefield.
The next generation of fMRI multimodal studies will determine the extent to which emotional responses and psychopharmaceutical impairment affect navigational task learning and performance. It is expected that some optimal level of emotional involvement in navigational decisions, especially those made under stress from high cognitive workload, sleep deprivation, or some induced affective disorder, can be determined; for example, it is likely that optimal decision making occurs somewhere between cold-as-ice detachment and “We’re all going to die!” paranoid neuroses. Such tools will be useful in training all categories of operators, whether tankers or snipers, or to provide feedback to helicopter or UCAV pilots, especially those who increasingly may view simulator training as a video game. Moreover, monitoring the efficiency of training and the effectiveness of alert systems can directly impact the risk of performance failure owing to interface problems and ineffective workload, as it has already been shown that pilots show different activation levels once they become fully trained in a maneuver (Peres et al., 2000), when they have reached their fastest reaction times (Mohamed et al., 2004), and when an alert interface is optimized to emphasize vigilance (Reilly, 2003). Thus, the basic groundwork is already in place to “watch soldier-operators think” during and after training that emphasizes attention to psychological states.
Generation-after-next fMRI imaging tools promise to deliver a capability for monitoring deep-brain emotional states in-flight. Research late in the next decade should also lead to simple paradigms and algorithms capable of benchmarking behavioral fitness in a matter of minutes as part of a weekly health regimen including examination of any neurochemistry changes in emotional reception and processing centers.
Once BOLD fMRI benchmarks are established, NIRS technology can be used to monitor cortical signals deemed important during actual task performance instead of a simulated paradigm. NIRS technology can also be used to monitor BOLD signals from increasing cerebral recruitment during extended dismounted activities or other long-duration mission-critical assignments (Drummond et al., 2004).
THE CURRENT OPERATIONAL STATE OF THE ART
Noncontact Multispectral Neurophysiological Sensors
Blood Pressure, Heart Rate, Interbeat Interval, Body Tremor (0.2 to 1.0 Hz), and Ballistocardiogram
Signal-processing algorithms (originally developed for underwater and spaceflight applications) have recently proved applicable to low-frequency improvement in conditions with a very high signal-to-noise ratio, such as for enabling real-time monitoring of numerous human neurological signals in both aware and unaware subjects. The signals illustrate embedded cardiovascular information. Such signal processing has been applied to the known technique of ballistocardiogram.
The validation of the subsystem, sensors embedded in a chair back, legs, and seat with RF signals “ported” to a remote station, was accomplished in a medical school: the subjects had arterial (radial) cannulation, and the software was subsequently “trained” by the ground-truth data. (This was, of course,
a straightforward series of statistical comparisons of signals of varying bandwidths, with the algorithm sequencing for “best fit.”) Electron paramagnetic resonance oxygen mappings (EPROMs) were constructed and devices “measured” (approximated by estimation) systolic and diastolic blood pressure with good accuracy (plus/minus 5.0 mm Hg) compared with that of mercury sphygmomanometers.
An interesting “fall-out” of the human subject validation experiments was that the heart rate and interbeat interval were accurate enough to calculate, from tables (programmed in the EPROM), near-real-time caloric expenditure.
Detection of Carotid Artery Blood Flow, Differential Stenosis, and Heart Rate
During the test and evaluation of a small, man-portable infrared (mercury-cadmium-telluride) detector it was observed that thermographic profiles of humans in the near field (>2 and <25 ft) could be obtained, overtly and covertly. The system was tested in a series of subjects and calibrated. The specificity and sensitivity to facial temperature bilaterally over the cheeks and forehead were seen to be operationally useful, and data were collected in less than 1 second (from a seated subject in a reasonably quiet pose) that were accurate to within 0.5 degrees C.
The data were compared and contrasted with breast-tumor and facial thermographic units being used clinically. The pre-stroke detection data were incorporated successfully into an adult male prestroke health-monitoring program and deployed for further operational testing and evaluation in a scenario identical to that in an interview/interrogation facility (e.g., used in operational settings). The systems were installed behind infrared-transparent, visually opaque mirrors. The rooms were constructed with low ambient temperatures (<65 degrees F) and fans to enhance the utility of clinical data collection.
Subjects were unaware of the data collection. Scans were read and interpreted by qualified radiologists for clinical use.
Covert Communication System, One-Way to Simulated Polygrapher
A system was developed and fabricated that used a very-low-power, highly collimated and directionally deployed CO2 laser to transmit voice data (1,000 to 10,000 Hz) to an operator’s external auditory meatus, with real-time questions asked for the purpose of interrogating an individual, outside, seated on an instrumented bench (as described above), but inaudibly to the subject. The system was tested and found to work exceptionally well, under reasonable ambient outdoor conditions.
Barkley, G.L. 2004. Controversies in neurophysiology. MEG is superior to EEG in localization of interictal epileptiform activity. Clinical Neurophysiology 115 (5): 1001-1009.
Calhoun, V.D., J.J. Pekar, and G.D. Pearlson. 2004. Alcohol intoxication effects on simulated driving: Exploring alcohol-dose effects on brain activation using functional MRI. Neuropsychopharmacology 29 (11): 2097-2117.
Chae, J.H., J. Jeong, B.S. Peterson, D.J. Kim, W.M. Bahk, T.Y. Jun, K.S. Kim, and K.Y. Kim. 2004. Dimensional complexity of the EEG in patients with posttraumatic stress disorder. Psychiatry Research 131 (1): 79-89.
Drummond, S.P., G.G. Brown, J.S. Salamat, and J.C. Gillin. 2004. Increasing task difficulty facilitates the cerebral compensatory response to total sleep deprivation. Sleep 27 (3): 445-451.
Genik II, Richard J., Christopher C. Green, Francis X. Graydon, and Robert E. Armstrong. 2005. Cognitive avionics and watching spaceflight crews think: Generation-after-next research tools in functional neuroimaging. Aviation, Space & Environmental Medicine, May, in press.
Graydon, Francis X., Richard Young, Mark D. Benton, Richard J. Genik II, Stefan Posse, Li Hsieh, and Christopher C. Green. 2004. Visual event detection during simulated driving: Identifying the neural correlates with functional neuroimaging. Transportation Research Part F: Traffic Psychology and Behavior 7 (4-5): 307-322.
Hamalainen, Matti, Riitta Hari, Risto J. Ilmoniemi, Jukka Knuutila, and Olli V. Lounasmaa. 1993. Magnetoencephalography—theory, instrumentation and applications to non-invasive studies of the working human brain. Review of Modern Physics 65 (2): 413-505.
Hatzinger, Martin, Ulrich M. Hemmeter, Serge Brand, Marcus Ising, and Edith Holsboer-Trachsler. 2004. Electroencephalographic sleep profiles in treatment course and long-term outcome of major depression: Association with DEX/CRH-test response. Journal of Psychiatric Research 38 (5): 453-465.
Jezzard, Peter, Paul M. Matthews, and Stephen M. Smith (eds.). 2003. Functional MRI: An Introduction to Methods. New York, NY: Oxford University Press.
Mohamed, Mona A., David M. Yousem, Aylin Tekes, Nina Browner, and Vince D. Calhoun. 2004. Correlation between the amplitude of cortical activation and reaction time: A functional MRI study. American Journal of Roentgenology 183 (3): 759-765.
Peres M., P.F. Van De Moortele, C. Pierard, S. Lehericy, P. Satabin, D. Le Bihan, and C.Y. Guezennec. 2000. Functional magnetic resonance imaging of mental strategy in a simulated aviation performance task. Aviation Space Environmental Medicine 71 (12): 1218-1231.
Ramos-Loyo, Julieta Andrés, Antonio Gonzalez-Garrido, Claudia Amezcua, and Miguel A. Guevara. 2004. Relationship between resting alpha activity and the ERPs obtained during a highly demanding selective attention task. International Journal of Psychophysiology 54 (3): 251-262.
Reilly, Clay E. 2003. Motor co-ordination in humans is guided by optimal feedback control. Journal of Neurology 250 (2): 257-258.
Uchiyama Y., K. Ebe, A. Kozato, T. Okada, and N. Sadato. 2003. The neural substrates of driving at a safe distance: A functional MRI study. Neuroscience Letters 352 (3): 199-202.