Biomagnetic studies of slow CNS disturbances. Gardner-Medwin AR (1993) In: Migraine: Basic Mechanisms and Treatment. ed. Lehmenkühler A. et al.. Urban & Schwarzenberg, Munchen, pp. 229-238

Biomagnetic studies of slow CNS disturbances
A.R. Gardner-Medwin
Department of Physiology,
University College London,
London WC1E 6BT,
UK

Running Head: MEG studies of slow CNS disturbances

INTRODUCTION

 Measurement of the magnetic fields produced by electrical events in the brain (magneto-encephalography: MEG) offers a non-invasive technique for study that has a number of advantages over the direct measurement of surface voltage changes (EEG). The advantage that has received the most study is its potential for localising the origin of a signal accurately. MEG signals are relatively little affected by the high resistance (bone and dura) encasing of the brain and the low resistance (scalp) outer tissue, which attenuates and distorts electric current flow to the sites of non-invasive electrodes. Much interest has centred round the localisation of epileptic foci, event related signals, and rhythmic activity. These signals mostly lie in the frequency band 1Hz-1kHz, for which MEG techniques nowadays tend to be best adapted.
 Migraine, particularly the aura of classical migraine, is known to be associated with some sort of pathophysiology of the brain, often specifically of parts of the cerebral cortex. The disturbances usually have a time course of minutes or hours. The visual aura in classical migraine has components (scintillating patterns at the region of onset of a scotoma) that suggest abnormal physiological activity at frequencies well above 1Hz, but the profound loss of function that spreads across the visual cortex in this form of migraine itself has a duration at any one site of several tens of seconds or minutes (e.g. Lashley, 1941). It is potentially of interest to look for MEG signals with this kind of time course under such conditions.
 Preliminary biomagnetic studies have been carried out in migraine patients, some of them using DC recording techniques (Barkley et al., 1990). There are formidable technical problems in such studies, due to the level of magnetic interference that is picked up at low frequencies even inside a shielded room, and the problems of artefact identification. This makes it difficult to establish what signals may be consistently related to specific aspects of the migraine syndrome, let alone where (in the brain or elsewhere) these signals may originate.
 One of the problems in migraine is that one is looking for magnetic signals that are not necessarily time locked in a consistent way to any other identifiable characteristic of the syndrome. They may also differ radically, because of different tissue geometry and structures involved, in different patients. They are therefore not amenable to averaging across episodes or across subjects, which can improve the signal-to-noise ratio in experimentally triggered disturbances such as spreading depression in animals (Gardner-Medwin et al. 1991, & Fig. 1). Satisfactory study of one-off slow signals requires a considerable technical improvement in signal-to-noise ratio and artefact identification.
 A particular reason for interest in slow magnetic signals in neurological conditions is that both ischemia and spreading depression are associated with very large and prolonged electro-physiological changes in affected regions: large negative extracellular potential changes, high potassium concentration, and large currents through glial cells involved in homeostasis (see Gardner-Medwin, 1981 for a discussion of these phenomena in relation to migraine). Spreading depression offers a good experimental model for studying the effects of such changes. The magnetic signals seen in anaesthetised rabbits during spreading depression consist of slow field shifts with a peak amplitude of about 1.5 pT approximately 400 s after induction of the disturbance with KCl application. The average inferred current dipole sources (Fig. 1B) have a similar time course, with variable rapid components at the onset and termination of KCl application. These signals were close to the limits of what could reliably be detected with conventional biomagnetic techniques using second order gradiometers (see below) in a screened room.

TECHNIQUES
 This paper sets out a novel approach to the improvement of biomagnetic signal-to-noise ratios at ultra low frequencies (0.001-0.1Hz). It is well known that to measure DC sources one must move the source relative to the magnetometer. Various approaches have been used, including comparison of fields with the subject under the magnetometer and at a remote distance either horizontally (Cohen & Kaufman, 1975) or vertically (Elbert, 1991) or with repeated scanning across the source with the subject on a moveable bed (Freake et al., 1988) or with an animal on a swing (Trontelj et al., 1989). The procedure employed here is to impose on the subject a constant horizontal circular motion under the magnetometer, with a radius of 12 mm and a period of normally 2-4 s for human subjects (Fig. 2). This is referred to as rotatory scanning (Gardner-Medwin, 1991). The subject is barely aware of this motion, provided it is smoothly executed.
 The rotatory scanning has the effect that any DC magnetic source on the moving plate produces an AC field at the frequency of rotation (Fig. 3). This AC field is proportional to the horizontal gradient of field due to the source, and is in a region of the spectrum where the noise amplitude in the critical units (pT/?Hz) is much less than at frequencies below 0.1Hz. The DC sources and fields can be inferred from measurements of the AC signals, rejecting completely the effects of DC field shifts from environmental interference.
 Automatic computer analysis of the signals allows measurements of the field gradients in two perpendicular directions to be derived in each cycle of the rotation. The directly measured fields can also of course be observed, and are usually averaged throughout each cycle. This provides significant data compaction, with still quite adequate time resolution for study of phenonena varying over tens of seconds or more. This gives a technique for essentially continuous high stability monitoring of DC fields, simultaneously on as many channels as the hardware and software will handle. Fig. 4 shows the simultaneous monitoring on 6 magnetic channels of a slowly changing current in a calibration dipole (period 20s, maximum peak to peak amplitude 24?A.mm, approximately as for the sources during spreading depression in Fig. 1). Substantial background field shifts took place during this 10min stretch of recording, larger than the directly measured sinusoidal signals (middle traces). These had only a transient and small effect on the gradient traces extracted from the rotatory scanning analysis and shown at the top (for one direction only). The data obtained wth the rotatory scanning techniques monitors faithfully the slow time course of the changes in the source.
 There are two incidental additional advantages that derive from the rotatory scanning technique:
  1. Most commercial magnetometers measure vertical or near vertical fields, which are zero directly over an active current dipole source. They have positive and negative maxima on either side of the dipole, as can be seen in Fig. 4 by noting the opposite phase of the large signals at the top and bottom of the mean directly measured fields. The field gradient on the other hand has a maximum directly over the dipole (as seen in the top traces in Fig. 4). This makes for somewhat more direct interpretation of the records in relation to dipole sources.
  2. Since the rotatory scanning derives horizontal gradients of the measured fields, it improves the relative sensitivity of the magnetometer to nearby sources rather than distant sources. Fields due to distant sources that are not on the moving table are totally rejected unless they happen to have a component of variation at the frequency of movement. Magnetic sources on the moving table, such as ferromagnetic contamination in the nose, teeth, lungs or abdomen, are however also sources of significant interference, especially if body movements occur. These contamination sources, though they may be very large, are mostly much more distant than the sources of interest in the brain. The signals they induce fall off as r-4 for an instrument designed as what is called a 1st order gradiometer and as r-5 for a 2nd order gradiometer (where r is the distance of the source, assumed to be a magnetic (not a current) dipole, of constant orientation). With these same instruments the field gradients measured with rotatory scanning fall off as r-5 and r-6 respectively. This can give a significant benefit. For the same reasons, a 2nd order gradiometer is generally better than a 1st order gradiometer at rejecting the effects of contamination.

VALIDATION
 Assessment of the rotatory scanning technique ideally requires a slowly changing physiological signal that is reproducible in man, with similar temporal characteristics to the signals of potential interest during brain pathologies. In collaboration with S. Swithenby and K. Fiaschi we are studying the field changes associated with the corneo-retinal current through the eye. This current rises slowly to about double its baseline level in the dark, reaching a peak approximately 8 minutes after the onset of light. The usual technique for monitoring these changes does not even involve DC electrical recording, but measures the rapid potential shift between the two sides of an eye (electro-oculogram) when a standard eye movement is made (Arden & Kelsey, 1962). Parallel measurements of magnetic shifts have also been made using standard eye movements (Katila et al., 1980). With care, the corneo-scleral DC potential differences can be measured directly (Nielsen XX, 1991). We are essentially carrying out the equivalent DC magnetic study, measuring directly the steady magnetic fields in the vicinity of the eye and their changes following a change in light level.
 Fig. 5 shows records obtained with a single channel magnetometer using rotatory, showing the field gradients on exposure to light after dark adaptation and also the simultaneous conventional measurements of the mean field over each cycle period. The sensor was positioned over the left temple. The gradients measured with the rotatory scanning analysis exhibit the characteristic time course of the changes in corneo-retinal current. Similar time courses following the onset of light were seen in the modulus of the gradient on each of 4 runs (with various sensor positions) on 2 subjects. Although these observations are very  preliminary and the changes have yet to be mapped satisfactorily, they demonstrate the advantage to be gained from the scanning technique. Note that the simultaneous direct field measurements in Fig. 5 show too many erratic DC shifts to be interpretable at all.

CONCLUSION
 Complete and confident characterisation of slow magnetic signals from a neurological patient necessarily requires that a great many factors combine at one time. Some of the trickiest of these are clinical factors: the patient must be able to produce the phenomena of interest within an acceptable time in the magnetic laboratory. He must also be sufficiently cooperative and comfortable to lie still for the time required. The kinds of improvement of signal to noise ratio described here should go a long way in improving the technical reliability of the records and reducing the stringency of the requirements that a good subject must be relatively free of magnetic contamination.

ACKNOWLEDGEMENT
This work is supported by the Wellcome Trust, and has in part depended on the collaboration and advice of S. Swithenby, K. Fiaschi, C. Guy, T. Elbert and their colleagues. Much of the equipment was made by D. Farquharson and P. Stukas.

REFERENCES

Arden G.B. & Kelsey J.H.: Changes produced by light in the standing potential of the human eye. J.Physiol. 151: 189-204 (1952)

Barkley G.L., Tepley N., Nagel-Leiby S., Moran J.E., Simkins D.O. & Welch, K.M.A.: Magnetoencephalographic studies of migraine. Headache 30: 428-434 (1990).

Cohen D., Kaufman L.A.: Magnetic determination of the relationship between the S-T segment shift and the injury current produced by coronary artery occlusion. Circulation Res. 35:414-424 (1974)

Elbert T., Braun C., Rockstroh B., Schneider S.: The measurement of tonic brain activity by means of magnetoencephalography. Proc. 8th Int'l. Conf. on Biomagnetism: 247-248 (1991)

Freake S.M., Swithenby S.J., Thomas I.M.: A miniature squid magnetometer for the detection of quasi-DC ionic current flow in developing organisms. In: Atsumi K, et al., eds. Biomagnetism '87, Tokyo Denki UP,Tokyo, 1988, pp. 434-437

Gardner-Medwin A.R.: Possible roles of vertebrate neuroglia in potassium dynamics, spreading depression and migraine. J. exp. Biol. 94:111-127 (1981)

Gardner-Medwin A.R.: Rotatory scanning for improved discrimination of very low frequency components of the magneto-encephalogram. Proc. 8th Int'l. Conf. on Biomagnetism: 51-52 (1991)

Gardner-Medwin A.R., Tepley N., Barkley G.L., Moran J., Nagel-Leiby S., Simkins R.T., Welch K.M.A.: Magnetic fields associated with spreading depression in anaesthetised rabbits. Brain Res 440:143-148 (1991)

Katila T., Maniewski R., Poutanen T., Varpula T. & Karp P.J.: Magnetic fields produced by the human eye. J. appl. Physics: 42:2454-2471 (1980)

Lashley K.S.: Patterns of cerebral integration indicated by the scotomas of migraine. Archs. Neurol. Psychiat. Chicago 45:331-339 (1941)

Nilsen XX In: Heckenlively J.R. & Arden G.B., eds.: Principles and practice of clinical electrophysiology of vision. Mosby Year Book, 1991, pp. XX

Trontelj Z., Pirnat J., Luznik J., Jazbinsek V., Valencic V., Krizaj D., Vodovnik L., Jercinovic A. : Measurement of Magnetic field near an acute surgical injury on the rabbit's thigh. In: Williamson SJ, et al., eds. Advances in Biomagnetism. Plenum Press, New York, 1989, pp. 417-420

Gardner-Medwin AR, Tepley N, Barkley GL, Moran J, Nagel-Leiby S, Simkins RT, Welch KMA. Brain Res 1991;440:143-148

Okada YC, Lauritzen M, Nicholson C. Brain Res 1989;442:184-190

Tepley N, Barkley GL, Moran J, Simkins RT, Welch KMA. In: Williamson SJ, et al., eds. Advances in Biomagnetism. New York: Plenum, 1990;327-330

Cohen D, Kaufman LA. Circulation Res 1974; 35:414-424

Freake SM, Swithenby SJ, Thomas IM. In: Atsumi K, et al., eds. Biomagnetism '87, Tokyo: Tokyo Denki UP, 1988;434-437

Katila TE. (1990) In: Williamson SJ, et al., eds. Advances in Biomagnetism. New York: Plenum, 1990;19-32
 
 

Fig. 1.  Biomagnetic changes recorded from anaesthetised rabbits following initiation of Spreading Depression (SD) with 150mM KCl applied at the cortical surface in a cup on the left hemisphere. A. Means and s.e.m. from the indicated numbers of recordings (N), of vertical field shifts at various recording sites above the heads of the rabbits at 400s after KCl application. B. Inferred net dipole source strength (mean ± s.e.m.) in the anterior posterior direction (0o) and rightward direction (90o) from ten recordings. Bar shows the time of KCl application. Data from Gardner-Medwin et al. (1991).

Fig. 2.  Apparatus used for constant rotatory scanning of a subject under a magnetometer. The subject stays in a constant orientation while each point moves in a 12mm diameter circle with a period of usually 2-4s. The drive is conveyed to the upper plate by a system of cords, pulleys, etc. engineered in non-magnetic materials. The (1st or 2nd order) magnetometer measures vertical or near vertical fields on any number of channels.

Fig. 3.  Fourier transform data with a steady calibration source of different strengths at a distance of 15mm below the magnetometer sensor coil. A signal at the frequency of rotatory scanning (ca. 0.9Hz) is superimposed on the background noise and is proportional to the source strength. Measurement of the Fourier components at the scanning frequency gives the horizontal gradients of the vertical field produced by DC sources that move with the table. Most of the background noise is excluded, as are gradients due to magnetic sources that do not move with the upper plate in Fig. 2. Data obtained with a 2nd order gradiometer in an unshielded room, in collaboration with S. Swithenby & K. Fiaschi (Open University, UK).

 Fig. 4.  Measurements of the field gradient in the X direction (upper traces) obtained with the rotatory scanning technique from 6 magnetometer channels ca. 48mm above a calibration dipole in the Y direction. Direct field measurements averaged over each scanning cycle are shown below (middle traces). Current in the dipole was varied as shown in the bottom trace (at ca. 0.05 Hz). The 6 channels were spaced by 20mm in the X direction, approximately symmetrically over the dipole. Note the shifts in field measurements due to DC shifts in external interference; these have only slight transient effects on gradients derived from rotatory scanning. Data obtained with 1st order gradiometers in a shielded room, in collaboration with T. Elbert (Munster, Germany).

Fig. 5.  Measurements of DC field changes following the onset of light adaptation in a human subject. The subject had been dark adapted for ca. 20min and was then exposed to ordinary room light. The head lay on its right side, with the left temporal aspect of the skull horizontal. The vertical field was measured ca. 22 mm above the temple and 40mm from the corner of the eye in the X direction (toe to head). Gradients in the X direction and Y direction (front to back) are shown, derived from rotatory scanning at 0.3 Hz. Simultaneous recording of the directly measured mean field (top trace) shows marked field shifts not reflected in the gradient measurements. The X and Y gradients show the characteristic slow transient time course of changes in the corneo-retinal current due to light adaptation, with a maximum about 10min after light onset. A single channel 2nd order gradiometer was used in an unshielded environment with active cancellation of the earth's field. Data obtained in collaboration with S. Swithenby & K. Fiaschi (Open University, UK).