Developmental Brain Research 154 (2005) 57 – 62 www.elsevier.com/locate/devbrainres
Research report
Fetal magnetoencephalography—a multimodal approach Hari Eswarana,*, Curtis L. Lowerya, James D. Wilsonb, Pam Murphya, Hubert Preissla,c a
Department of Obstetrics and Gynecology, University of Arkansas for Medical Sciences, 4301 W. Markham, Slot 518, Little Rock, Arkansas 72205, United States b Graduate Institute of Technology, University of Arkansas at Little Rock, Little Rock, Arkansas, United States c MEG Center, University of Tuebingen, Tuebingen, Germany Accepted 5 October 2004 Available online 5 November 2004
Abstract Past studies have shown the feasibility of recording fetal evoked responses to external stimuli using a non-invasive technique called magnetoencephalography (MEG). These studies were all performed using either auditory or visual stimuli and showed a fairly low detection rate for each modality, thus making this technology currently unreliable for fetal clinical applications. This study is based on the hypothesis that a multimodal approach of applying both auditory and visual stimulation paradigms in successive recording sessions could improve the probability of obtaining a fetal evoked response. A total of 34 studies were performed on 11 normal healthy fetuses at different stages of gestation starting as early as 28 weeks with a 151-channel fetal MEG system. The success rate of obtaining a response to either (or both) stimuli from a study at a given gestation age was 91%. All the 11 fetuses showed a response at least once over the gestation period the recordings were performed. A multimodal testing approach can improve the ability of the MEG technique to reliably monitor the functional development of the fetal brain. D 2004 Elsevier B.V. All rights reserved. Keywords: Magnetoencephalography; Fetal auditory evoked response; Fetal visual evoked response; Brain activity; Fetal neurological assessment
1. Introduction Fetal electrocortical activity was first recorded noninvasively by Lindsey [10] in 1942 using a transabdominal approach. Technological limitations of the time and general problems related to transabdominal electrical recordings resulted in poor quality signals. In 1969, Rosen and Scibetta [15] and Scibetta et al. [18] began recording electrocortical activity from fetuses during labor, using electrodes placed directly on the fetal scalp after the rupture of the amniotic membranes. In an effort to find a non-invasive and efficient fetal monitoring technique, Kariniemi et al. [7] explored an approach using biomagnetic principles to study fetal activity. The technique of recording magnetic signals from the brain is called magnetoencephalography (MEG). MEG
* Corresponding author. Tel.: +1 501 526 4334; fax: +1 501 603 1544. E-mail address:
[email protected] (H. Eswaran). 0165-3806/$ - see front matter D 2004 Elsevier B.V. All rights reserved. doi:10.1016/j.devbrainres.2004.10.003
records magnetic signals corresponding to electrical currents in biological tissue [12]. In contrast to direct electrical recording of the currents, the magnetic signals are not distorted by different layering of biological tissues. Based on this, MEG is the only non-invasive method, which allows the recording of magnetic fields generated in the fetal brain in utero. In the past decade, this technology has mostly been applied to magnetism associated with the adult brain and neuromuscular system [11]. In fact, many studies [6,12,14] have been conducted on human adult magnetic sensory evoked fields. However, this technology has not been fully exploited for fetal monitoring although the magnetic field of the fetal heart was detected as early as 1974 [7]. In 1985, Blum et al. [1] reported for the first time the successful recording of fetal MEG (fMEG). An ultrasound recording enabled them to locate the position of the fetal skull, position a one-channel neuromagnetometer over the maternal abdomen and record fetal auditory evoked responses
58
H. Eswaran et al. / Developmental Brain Research 154 (2005) 57–62
(AER). In their study, the duration of the sound stimulus was 100 ms and the intensity varied from 40 to 100 dB (HL) to allow fetal and maternal adaptation. Postnatal recordings of the subject showed similar brain signals compared with the prenatal observations. More recently, Wakai et al. [21] conducted long latency fetal AER studies on 14 pregnant women with gestational ages ranging from 36 to 40 weeks. In this study, a seven-channel, second order SQUID gradiometer was used to acquire the fetal brain activity. The waveform component corresponding to the adult N100 appears to be delayed significantly in the fetus and neonate. Various fMEG investigators (including our group) have recorded a peak auditory evoked field amplitude ranging from approximately 30 to 175 fT and latency of the primary response component from 125 to over 200 ms [2,3,8, 9,13,17,21,22]. In order to monitor the neurological status of the fetus, we conducted serial recordings (starting at 28 weeks of gestation) of the fetal auditory evoked response and visual evoked response using the SARA (SQUID Array for Reproductive Assessment) system. SARA is a fMEG device specifically designed to study multiple aspects of maternal and fetal physiology. Most fMEG studies have primarily focused on recording cortical evoked responses to auditory stimuli [2,3,8,9,13,17,21,22]. All of these studies, including our own [3,13], have shown the detection rate of AERs to be around 50 %. Further investigation by our group [3] revealed that multiple recordings occurring over a short period of time in healthy fetuses increased the detection rate of fetal evoked response to 80%. Eswaran et al. [4] have also demonstrated the feasibility of recording visual evoked responses on the fetus. In this pilot study with 10 subjects, the detection rate of VERs was around 40%. A further study [5] with a larger population showed an improved detection rate of 58%. Despite the fact the detection rate of each modality is still low, the current technology provides us an opportunity to investigate the fetal brain function using multimodal approach. Our research hypothesis is that the success rate of obtaining an evoked response could be improved by using a multimodal
approach of applying the visual stimulus paradigm in conjunction with auditory paradigm. In this paper, we present and discuss the results from a study on fetuses who were presented with auditory and visual stimuli in separate but successive recording sessions.
2. Methods 2.1. Instrument The MEG recordings were performed using the 151 sensor SARA system (Eswaran, 2002a,b) installed at University of Arkansas for Medical Sciences (UAMS). During the recording session, the mother sits and leans forward against the smooth surface of the curved array, thus allowing the sensors to receive signals from the entire maternal abdomen (Fig. 1). SARA is installed in a magnetically shielded room (Vakuumschmelze, Germany), to reduce the effects of environmental noise. 2.2. Subjects and recording protocol A total of 34 studies (68 recordings—34 auditory and 34 visual) were performed on 11 healthy fetuses at different stages of gestation starting as early as 28 weeks. The Institutional Review Board approved research protocol was discussed with each patient and a written consent was obtained. After the first recording, the patients were asked to return for follow-up recordings at 2-week intervals. Prior to the MEG measurements, an ultrasound was recorded on all the fetuses. The MEG data were recorded in a continuous mode at a sampling rate of 312.5 Hz. The recording session lasted for 6 min for each stimulus modality-first visual followed by auditory. 2.3. Visual stimulus A flash stimulus was used to elicit the visual evoked response. The light source, located outside the shielded
Fig. 1. (Left) A subject positioned on the SARA system for recording. (Right) Layout of the 151 sensor array over the maternal abdomen.
H. Eswaran et al. / Developmental Brain Research 154 (2005) 57–62
room, utilizes a light emitting diode (LED) array (Opto Technology model OTL630A-5-10-66-E, wavelength—625 nm). A 77-m long fiber-optic cable channels the light flash from the LED array (outside the shielded room) to the maternal abdomen. The LED array emits light at wavelength of 629 nm, which is in the visible range (red light). The peak illuminance at the exit of the fiber-optic cable of the light pulse was 8800 lx. This light pulse delivered to the maternal abdomen was considered safe for the fetus since it is of short duration, contains no short wavelength radiation, and has an intensity much lower than sunlight on a bright day (approximately 100,000 lx). The duration of the flash was 33 ms with a interstimulus interval of 2 s (F250 ms randomization). 2.4. Auditory stimulus The auditory stimulus system consisted of a speaker which was situated outside the shielded room and attached to 12 ft of plastic tubing. The tubing was then passed through a small opening into the shielded room and distally connected to an air-filled bag. This bag was secured with an elastic belt over the apex of the maternal abdomen without increasing the distance between the sensors and fetal head. The sound stimulus was a 1-s tone burst with a combination 500 Hz (frequent tone—80%) and 1 kHz (rare tone—20%) frequency. The rare tone was used to minimize any possible habituation to the frequent tone by the fetus. The intensity of the tone recorded at the maternal abdomen was 120 dB and the tones were delivered at an interstimulus interval of 2 s (F500 ms randomization).
3. Results A total of 68 recordings (34 visual and 34 auditory) were analyzed. Table 1 shows the gestational age of the recorded fetuses along with their visual and auditory evoked response latency values for the highest observed peak and the corresponding amplitudes. All the 11 fetuses showed a response at least once over the gestation period the recordings were performed. The average latency and amplitude of response to visual modality was 254.7 ms and 26 fT, respectively. Similarly for the auditory response, the average latency and amplitude was 264.6 ms and 30.1 fT, respectively. Fig. 2 shows the visual (top) and auditory (bottom) evoked response from the same fetus at 31 and 33 weeks of gestation respectively. This figure also shows a color coded map depicting the location of magnetic activity over the 151 sensor array montage. The color-coding is based on the intensity and the direction of the magnetic fields at a given time point. The maps in Fig. 2 show the
Table 1 Results of AER and VER studies performed with SARA Patient number
Gestational age (weeks)
201
28 29 31 33 35 28 30 32 34 31 33 37 27 30 36 29 32 34 30 32 34 27 29 31 36 32 34 36 38 34 30 34 29 33
202
2.5. Data analysis Interfering maternal and fetal heart signal components were removed by applying a spatial-filter based projection algorithm [16,19,20]. This method is first applied on the raw signal to remove the maternal heart artifact and then applied again in the resulting data set (devoid of maternal signal) to remove fetal heart artifact. The evoked responses were obtained by averaging the extracted time slices from the continuous data set starting at 0.2 s before the onset of the stimulus until 0.8 s after the onset of the stimulus. In the case of auditory stimulation, the averages were computed only for the frequent tones. The results were then bandpass filtered between 0.5 and 10 Hz. A response was considered good if at least one peak was observed in the averaged response window. The differences detected between no response and discernible response were determined based on a method called plus–minus averaging. The mathematical comparison between the conventional averaging and the plus–minus averaging gives a measure of signal to noise ratio, the amount of noise present and thus the reliability of the signals. Two conditions that were used for marking bdiscernible responseQ were: (1) the highest latency peak should be greater than 8fT and (2) the signal to noise ratio was at least 2:1.
59
203
204
205
206
207
208
209 211 212 a
Visual stimulus
Auditory stimulus
Latencya (ms)
Amplitude (fT)
Latencya (ms)
Amplitude (fT)
323.2 384
30.8 31.4
300.8 182 384 323.3
37.3 38.1 11.7 24
387.2
77.5
262.4 374
78.1 35.3
140 313.6 272
18.5 19.4 19.3
190 387 240 140 252 236 396 392.6
30.9 18.4 24 91 45.2 33.8 21 29.7
393.6
13.5
233 176
15.8 25.5
153.6 185.6
24.5 11.5
236.8 144 217.6 208
15 23.9 11 10.9
153 316.8 147 236.8 150
18.5 60 17.4 9.1 9
371
12.1
336 206
27.8 17.5
284 176 304 163
30.2 25.6 29.6 20
304
17.6
220.8 217.6
43.4 25.7
Latency and amplitude of the most prominent peak observed.
60
H. Eswaran et al. / Developmental Brain Research 154 (2005) 57–62
Fig. 2. (Top) Visual evoked response from a fetus at 31 weeks of gestation. (Bottom) Auditory evoked response from a fetus at 33 weeks of gestation. A colorcoded map shows the location of the corresponding magnetic activity over the 151 sensor layout. The maps show the location of magnetic activity corresponding to a latency of 263.4 ms for VER (top right) and latency of 289.4 ms for AER (top left).
location of magnetic activity corresponding to a latency of 263.4 ms for VER (top right) and latency of 289.4 ms for AER (top left). The generating source would normally be located between maxima of the opposite polarities. In both the cases, the location of this peak activity corresponded
with position of the fetal head recorded by ultrasound prior to MEG recordings. Fig. 3 shows a representative VER from recordings of a single fetus at 27 (right) and 36 (left) weeks of gestation. The peak latencies can be observed at 185 ms and 316 ms
Fig. 3. Representative fetal VER from a single fetus at 27 (right) and 36 (left) weeks of gestation. The peak latencies can be observed at 185 and 316 ms for the 27 and 36 gestation weeks, respectively.
H. Eswaran et al. / Developmental Brain Research 154 (2005) 57–62
61
Fig. 4. Representative AER response from single fetus recorded at 32 (right) and 34 (left) weeks of gestation. The peak latencies can be observed at 236 and 144 ms for the 32 and 34 gestation weeks, respectively.
for the 27 and 36 gestation weeks, respectively. Similarly, Fig. 4 depicts serially recorded AERs from a fetus at 32 weeks (right) with latency peak at 236 ms (left) and 34 weeks with a latency peak at 144 ms. It was found that 23 of the 34 (68%) recordings showed a discernible response to visual stimuli. In the case of auditory stimuli, 21 out of 34 (62%) recordings showed discernible responses. In terms of obtaining a response to either (or both) stimuli during a study at a given gestation age, the success rate was 91% (31/34 studies). There were 13 studies where responses were obtained for both modalities.
4. Discussion The results of this study show that it is possible to record both auditory and visual evoked responses from the fetus as early as 28 weeks of gestation. As shown in the results, the response to each modality ranged from 62% to 68%. However, by defining the success rate as a positive response to either auditory or visual stimuli, this value was found to be much higher at 91%. From this study, it is clear that a single testing modality would not be sufficient to reliably assess the fetal brain activity and that a multimodal testing approach would provide a means of more complete assessment of the functional development of the brain. The relatively low success rate reported by various fMEG investigators [2–4,8,9,13,17,21,22] is attributed to several factors, including the distance between the fetal head and the sensors, movement of the fetal head during data acquisition, sub-optimal placement of the sensors relative to the abdominal region containing most of the evoked signal, and the gestational age and neurological maturity of the fetal brain. As is evident, some of these factors cannot be controlled and will vary among subjects. In a previous study [3] by our group, a preliminary analysis of the correlation between the fetal head position, orientation or distance from
maternal skin and the successful detection of evoked response was not very conclusive. But if we accept the premise that the absence of an observable evoked response in the fetus could be an indicator for fetal neurological impairment, we need to assure that the fMEG technique does not suffer from a high false positive rate. However, a low success rate of detection for each stimulus modality by itself, leads to an unacceptably high false-positive rate that could preclude its use. A multimodal approach that includes the assessment of auditory and visual evoked responses and perhaps other indicators of fetal status such as spontaneous brain activity and fetal heart could possibly reduce the number of false-positives. In other words, if evoked responses are dependent on certain conditions such as fetal state or position, it may be still be possible to obtain either AER or VER in the normal fetus through serial recordings over the gestation period since these responses should be observable once they are developed. The above results indicate that we were able to obtain a response from the normal fetuses at least once during their gestation period. In the case of the fetus with impaired brain function, observing either AER or VER would be unlikely despite repeated measurements over gestation. In terms of latency values, it can be observed from the table that there is a large spread in this value across the gestation ages. This can be explained by the fact that these latency values were recorded from the most prominent peaks which do not necessarily correspond to same latency component across all subjects or gestational ages. It is possible that, depending on the fetal head position incident on sensor array (which cannot be controlled), one latency component is dominant over the other. Since detectability of the response was the focus of the study, only the main prominent peak was chosen to accomplish this task. One goal is to develop MEG into a useful screening tool for detection of at least the severely neurologically damaged fetus. However, ultimately, it may be feasible to perform
62
H. Eswaran et al. / Developmental Brain Research 154 (2005) 57–62
serial multimodal fMEG exams on the high risk fetus in order to detect the early onset of neurological damage. If we can reliably detect early fetal neurological impairment with this approach, interventions could be developed to improve outcome. Additional studies are currently in progress to establish the clinical viability of fetal MEG.
Acknowledgements This work is supported by the grant from the National Institute of Neurological Diseases and Stroke (5R01NS36277) and National Institute of Biomedical Imaging and Bioengineering (1R33EB00987), National Institutes of Health. The authors wish to acknowledge the support of Dr. Stephen E. Robinson and Dr. Jiri Vrba, CTF Systems (subsidiary of VSM MedTech.), Canada.
References [1] T. Blum, E. Saling, R. Bauer, First magnetoencephalographic recordings of the brain activity of the human fetus, Br. J. Obstet. Gynaecol. 92 (12) (1985) 1224 – 1229. [2] H. Eswaran, C.L. Lowery, S.E. Robinson, J.D. Wilson, D. Cheyne, D. McKenzie, Challenges of recording human fetal auditory evoked response using magnetoencephalography, J. Matern.-Fetal Med. 9 (5) (2000) 303 – 307. [3] H. Eswaran, J.D. Wilson, H. Preissl, S.E. Robinson, P. Murphy, D.F. Rose, J. Vrba, C.L. Lowery, Short-term serial magnetoencephalographic recordings of fetal auditory evoked responses, Neurosci. Lett. 31 (2) (2002) 128 – 132. [4] H. Eswaran, J.D. Wilson, H. Preissl, S.E. Robinson, P. Murphy, D.F. Rose, J. Vrba, C.L. Lowery, Magnetoencephalographic recordings of visual evoked brain activity in the human fetus, The Lancet 360 (9335) (2002) 779 – 780. [5] H. Eswaran, C.L. Lowery, J.D. Wilson, P. Murphy, H. Preissl, Functional development of the visual system in human fetus using magnetoencephalography, Exp. Neurol. (2004 Aug.) (in press). [6] R. Hari, K. Aittoniemi, M.L. Jarvinen, T. Katila, T. Varpula, Auditory evoked transient and sustained magnetic fields of the human brain, Exp. Brain Res. 4 (1980) 237 – 240. [7] V. Kariniemi, J. Ahophelto, P.J. Karp, T.E. Katlia, The fetal magnetocardiogram, J. Perinat. Med. 2 (1974) 214.
[8] A. Kotini, N. Koutlaki, P. Anninos, V. Liberis, A. Hatzimihael, P. Anastasiadis, Fetal magnetoencephalography in intrauterine growth retarded pregnancies, Prenat. Diagn. 22 (2002) 1093 – 1097. [9] J.M. Lengle, M. Chen, R.T. Wakai, Improved neuromagnetic detection of fetal and neonatal auditory evoked responses, Clin. Neurophysiol. 112 (2000) 785 – 792. [10] D.B. Lindsey, Head and brain potentials of human fetuses in utero, Am. J. Psychol. 55 (1942). [11] O.V. Lounasmaa, M. Hamalainen, R. Hari, R. Salmelin, Information processing in the human brain: magnetoencephalographic approach, [Review], Proc. Natl. Acad. Sci. U.S.A., vol. 93, 1996, pp. 8809 – 8815. [12] J. Malmivuo, R. Plonsey, Bioelectromagnetism, Oxford University Press, New York, 1995. [13] H. Preissl, H. Eswaran, J.D. Wilson, S.E. Robinson, J. Vrba, P. Murphy, C.L. Lowery, Redefining fetal evoked fields with biomagnetic recordings over the whole maternal abdomen, Proceedings of 23rd Annual International Conference of the IEEE Engineering Medicine and Biology, IEEE Press, 2001. [14] M. Riete, J. Eldrich, J.T. Zimmermann, J.E. Zimmermann, Human magnetic auditory evoked fields, Electroencephalogr. Clin. Neurophysiol. 45 (1978) 114 – 117. [15] M.G. Rosen, J.J. Scibetta, The human fetal electroencephalogram: I. An electrode for continuous recording during labor, Am. J. Obstet. Gynecol. 104 (1969) 1057 – 1060. [16] M. Samonas, M. Petrou, A.A. Ioannides, Identification and elimination of cardiac contribution in single-trial magnetoencephalographic signals, IEEE Trans. Biomed. Eng. 44 (4) (1997) 350 – 386. [17] U. Schneider, E. Schleussner, J. Haueisen, H. Nowak, H.J. Seewald, Signal analysis of auditory evoked cortical fields in fetal magnetoencephalography, Brain Topogr. 14 (1) (2000) 69 – 79. [18] J.J. Scibetta, M.G. Rosen, C.J. Hochberg, L. Chik, Human fetal brain response to sound during labor, Am. J. Obstet. Gynecol. 109 (1971) 82 – 85. [19] M.A. Uusitalo, R.J. Ilmoniemi, Signal-space projection method for separating MEG or EEG into components, Med. Biol. Eng. Comput. 35 (2) (1997) 135 – 140. [20] J. Vrba, S.E. Robinson, J. McCubbin, C.L. Lowery, H. Preissl, H. Eswaran, J.D. Wilson, P. Murphy, Fetal MEG redistribution by projection operators, IEEE Trans. Biomed. Eng. (2004 July) (in press). [21] R.T. Wakai, A.C. Leuthold, C.B. Martin, Fetal auditory evoked responses detected by magnetoencephalography, Am. J. Obstet. Gynecol. 174 (5) (1996) 1484 – 1486. [22] F. Zappasodi, F. Tecchio, V. Pizzella, C. Emanuele, G.V. Romano, G. Filligoi, P.M. Rossini, Detection of fetal auditory evoked responses by means of magnetoencephalography, Brain Res. 917 (2) (2001) 167 – 173.
