Mapping cortical haemodynamics during neonatal seizures using diffuse optical tomography: a case study

Abstract

Seizures in the newborn brain represent a major challenge to neonatal medicine. Neonatal seizures are poorly classified, under-diagnosed, difficult to treat and are associated with poor neurodevelopmental outcome. Video-EEG is the current gold-standard approach for seizure detection and monitoring. Interpreting neonatal EEG requires expertise and the impact of seizures on the developing brain remains poorly understood. In this case study we present the first ever images of the haemodynamic impact of seizures on the human infant brain, obtained using simultaneous diffuse optical tomography (DOT) and video-EEG with whole-scalp coverage. Seven discrete periods of ictal electrographic activity were observed during a 60 minute recording of an infant with hypoxic-ischaemic encephalopathy. The resulting DOT images show a remarkably consistent, high-amplitude, biphasic pattern of changes in cortical blood volume and oxygenation in response to each electrographic event. While there is spatial variation across the cortex, the dominant haemodynamic response to seizure activity consists of an initial increase in cortical blood volume prior to a large and extended decrease typically lasting several minutes. This case study demonstrates the wealth of physiologically and clinically relevant information that DOT-EEG techniques can yield. The consistency and scale of the haemodynamic responses observed here also suggest that DOT-EEG has the potential to provide improved detection of neonatal seizures.

Keywords: Diffuse optical tomography (DOT); Functional near infrared spectroscopy (fNIRS); Hypoxic–ischaemic encephalopathy (HIE); Neonatal seizures.

Figures

Fig. 1

Panel a) shows a two-dimensional representation of the array design including the positions of the sources, detectors and the EEG electrodes. Panel b) provides an example frame from the video-EEG and panel c) shows a photograph of the DOT–EEG array on an infant's head.

Fig. 2

Bipolar EEG data showing the beginning of event 3 at 1442.5 s. A burst and a subsequent period of suppression are apparent in the 30 s leading up to the onset of the electrographic seizure.

Fig. 3

The raw DOT intensity data are shown in panel a). Panel b) shows the synchronized bipolar EEG data. Note the distinct periods of electrographic hyperactivity which are clearly visible despite the discontinuous nature of the EEG. Also note the dramatic changes in optical intensity which are temporally correlated with the EEG events.

Fig. 4

The average haemodynamic concentration changes (relative to their mean) for the hour of recording. Panel a) shows the data, high-pass filtered at 1 Hz, averaged across all good channels. The grey shaded areas show the onset and duration of the clinically identified electrographic seizures. All three HbO, HbR and HbT signals exhibit consistent behaviour across events. The events are consistently associated with an increase in HbO, HbR and HbT followed by an extended decrease below baseline. The linearly detrended version of this data is shown in panel b) to highlight the difficulty in determining the onset of the haemodynamic changes.

Fig. 5

The variations in HbT for two specific channels over the left frontal lobe, relative to the mean value for the hour of acquisition. The array layout is shown to indicate the location of the two channels. The seizures are indicated by the grey shaded periods. Despite their close proximity, the channels present very different haemodynamic responses to the electrographic events. The channel shown in the upper panel (a) is dominated by increases in HbT relative to baseline while the channel shown in the lower panel (b) is dominated by decreases in HbT.

Fig. 6

Reconstructed images of the changes in HbT associated with each of the 7 identified seizures events for 30 s prior to the electrographic onset, for the point of onset, for the time at which the spatially averaged HbT signal reaches a maximum, for the midpoint between the spatially averaged HbT maximum and the subsequent minimum, for the minimum and for the point at which the spatially averaged signal appears to have recovered to a stable state. These images are changes in HbT relative to a baseline defined as the mean of the period between 60 and 30 s prior to the electrographic seizure onset.

Fig. 7

A sequence of images showing the changes in HbT associated with event 3. The upper axes show the changes in haemoglobin concentration spatially averaged across the grey matter surface. Seven distinct time-points are identified and the associate reconstructed images of the changes in HbT concentration are shown in dorsal and left and right lateral views. All data are changes relative to a baseline defined as the mean of the period between 60 and 30 s prior to the electrographic seizure onset.