Altered functional connectivity in preterm neonates with intraventricular hemorrhage assessed using functional near-infrared spectroscopy

Abstract

Intraventricular hemorrhage (IVH) is a common neurological injury following very preterm birth. Resting-state functional connectivity (RSFC) using functional magnetic resonance imaging (fMRI) is associated with injury severity; yet, fMRI is impractical for use in intensive care settings. Functional near-infrared spectroscopy (fNIRS) measures RSFC through cerebral hemodynamics and has greater bedside accessibility than fMRI. We evaluated RSFC in preterm neonates with IVH using fNIRS and fMRI at term-equivalent age, and compared fNIRS connectivity between healthy newborns and those with IVH. Sixteen very preterm born neonates were scanned with fMRI and fNIRS. Additionally, fifteen healthy newborns were scanned with fNIRS. In preterms with IVH, fNIRS and fMRI connectivity maps were compared using Euclidean and Jaccard distances. The severity of IVH in relation to fNIRS-RSFC strength was examined using generalized linear models. fNIRS and fMRI RSFC maps showed good correspondence. Connectivity strength was significantly lower in healthy newborns (p-value = 0.023) and preterm infants with mild IVH (p-value = 0.026) compared to infants with moderate/severe IVH. fNIRS has potential to be a new bedside tool for assessing brain injury and monitoring cerebral hemodynamics, as well as a promising biomarker for IVH severity in very preterm born infants.

Keywords: Functional connectivity; Functional near-infrared spectroscopy; Intraventricular hemorrhage.

Fig. 1

Examples of (a) a 20-by-20 connectivity map and (b) an 8-by-8 map for the signals extracted from the fNIRS channels in the left and right hemispheres from the temporal, parietal, frontal and occipital lobes. The 8-by-8 map was obtained by averaging fNIRS signals from one lobe then correlating with each other. (c) Lobe-based analysis of the fMRI data. A hemisphere was sectioned into temporal, parietal, frontal and occipital lobes. (d) Based on the section, a connectivity map was built from the fMRI image using averaged BOLD series from each region. All figures were developed from the data from the same participant in the IVH group.

Fig. 2

Similarity of RSFC maps between fNIRS and fMRI with sparsity, measured with Euclidean and Jaccard distances on weighted and binarized maps, respectively. Means and confidence intervals (CI) were estimated using a bootstrapping method.

Fig. 3

Lobe-wise similarity maps at sparsity of 38%. Higher values reflect greater similarity between the two modalities. Average values among channels were 0.55 (upper left, standard deviation [SD] 0.16, CI 0.48–0.61), 0.66 (lower left, SD 0.18, CI 0.59–0.73), 0.49 (upper right, SD 0.16, CI 0.43–0.55) and 0.63 (lower right, SD 0.19, CI 0.56–0.71).

Fig. 4

Very preterm born neonates with moderate/severe IVH scanned at term-equivalent age showed increased HbO (left) and HHb (right) connectivity values relative to healthy newborns and neonates with mild IVH. **p < 0.01, *p < 0.05.