The link between Near Infrared Spectroscopy (NIRS) and EEG baseline signals in neonates : a possible clinical routine application for the NIRS technique.

1GRAMFC, Biophysics & Neurophysiology Lab., Faculty of Medicine, Amiens, France.2Nervous system functional explorations, North Hospital, Amiens, France.3Pediatric Intensive Care Unit, North Hospital, Amiens, France.

N. Roche-Labarbe1, A. Aarabi2, G. Kongolo3, R. Grebe1, and F. Wallois1,2

GRAMFCGroupe de Recherche sur l'Analyse Multimodale de la Fonction Cérébrale

* nadege.roche@u-picardie.fr

Introduction

Methods Results

Conclusion

- simultaneous 10-20 EEG (Deltamed Coherence® system) and left temporal NIRS (frequency-domain spectrophotometer Imagent®, ISS Inc.) recording

- 6 healthy premature neonates (gestational age = 29 to 33 weeks at measurement) and 3 premature neonates presenting a background of neurological suffering but normal EEG at that time (gestational age = 30 to 35 weeks at measurement)

In premature neonates, the Electroencephalography shows a physiological discontinuity of electrical activity during quiet sleep (Lamblin et al. 1999). On the other hand, Near Infrared Spectroscopy shows spontaneous oscillations of hemoglobin oxygenation. Similar oscillations are visible in term neonates (Taga et al. 2000) and adults (Hoshi et al. 1998) with NIRS, and with other functional imaging techniques (Obrig et al. 2000), but it had primarily been described as the result of vasomotion and considered as a physiological artifact of limited interest.

Until now, the origin of these oscillations and their possible relations with neuronal activity have not been established. Here we show that the bursts of electroencephalographic activity in neonates during quiet sleep are coupled to a transient stereotyped haemodynamic response.

Our results indicate that premature infants from 28 weeks of gestational age are capable of neurovascular coupling, and our data on pathological cases suggest that a neurological suffering can modify its pattern. If confirmed by a larger number of cases, this pattern could be a predictive parameter, and become a clinical application for the NIRS technique.

Besides, we now propose that the alternation of EEG hypoactive and active periods in neonates during quiet sleep corresponds to a metabolic activation, and is an energy-consuming mechanism.

We expect that this work will be the first of a series studying the relationships between the electrical and haemodynamic systems in the infants brain, especially physiological activities and not only induced or pathological responses.

Concomitant acquisition of EEG with NIRS provides a new insight into cerebral function in neonates, that can significantly improve our understanding of both signals and facilitate the integration of near infrared spectroscopy into the set of existing tools in Neurology.

Ref.Lamblin, M. D., M. André, et al. (1999). Neurophysiologie Clinique 29: 123-219.

Taga, G., Y. Konishi, et al. (2000). Neuroscience Letters 282: 101-104Hoshi, Y., S. Kosaka, et al. (1998). Neuroscience Letters 245: 147-150.

Obrig, H., M. Neufang, et al. (2000). NeuroImage 12: 623-639.

677T-AM

A. Example of EEG-NIRS recording.

B. Example of simultaneous EEG-NIRS acquisition (1 minute). On NIRS graph, solid line is [HbO], dashed line is [Hb].

AB

We selected 20 EEG bursts (Fig. 2) in quiet sleep for each patient and extracted the corresponding HbO and Hb concentrations traces from NIRS recordings.

Example of selected EEG bursts (on a 20 s EEG page).

Vertical black lines represent the selected bursts onset.

ANOVA for repeated measurements showed for normal patients a highly significant effect (p < 0.001) of the burst onset [HbO], and no statistical difference between patients. Including thepathological cases, we observe a difference (p < 0.05) with normal subjects. Effects on [Hb] were never significant. Newman-Keuls post-hoc comparisons for all patients highlighted the effect of the EEG bursts onset: a significant decrease of [HbO] values from 1 to 5 s after the onset, followed by a significant increase from 7 to 15 s after (p < 0.05).

NIRS average traces for a normal patient.

Concentrations are given +/- confidence interval 95%. The vertical line represents the EEG bursts onset.

Grand averages were calculated for normal and pathological subjects. The pathological subjects patterns show approximately the same behavior than the normal, but the initial deoxygenation appears much stronger.

Black lines are the grand average of normal patients, gray lines of the pathological subjects. Solid lines are HbO, dashed lines are Hb. The vertical line represents the EEG bursts onset.