Presented at the Neonatal Society 2016 Summer Meeting.
Chalia M1,2, Cooper RJ1,3, Lee CW1,2, Dempsey L1,3, Hebden JC1,3, Austin T1,2
1 neoLAB, The Evelyn Perinatal Imaging Centre, Cambridge University Hospitals NHS Foundation Trust
2 Department of Neonatology, Cambridge University Hospitals NHS Foundation Trust
3 Department of Medical Physics and Biomedical Engineering, UCL
Background: PAIS occurs in 1:2500 infants. It is secondary to occlusion of a large cerebral artery and potentially results in hemiplegic cerebral palsy, cognitive impairment and epilepsy in childhood. Clinical management in the neonatal period is restricted to symptomatic treatment; a key contributing factor to this is the lack of investigating tools for rapid detection of PAIS. Brain MRI remains the gold standard in diagnosing PAIS, but logistically it cannot always occur in the time-critical acute phase. Diffuse Optical Tomography (DOT) is a safe, noninvasive neuro-monitoring tool, which allows changes in cerebral haemoglobin concentrations to be imaged at the cot-side. Low-frequency oscillations in cerebral haemoglobin concentrations are known to be associated with cerebral autoregulation processes (1) and ‘resting-state’ functional brain networks (2). This study seeks to test the hypothesis that the power spectra of these LFOs differs between the healthy and PAIS-affected regions of the infant brain.
Methods: As part of a larger study, we have recruited healthy term infants as controls and infants who presented with seizures in the first 72 hours of life (3). These infants were scanned using the UCL Optical Imaging System for 1-2 hours. We retrospectively analysed data from 4 patients diagnosed with PAIS on MRI, 4 patients with no cortical changes or lateralised findings on MRI and 3 healthy term infants. Following manual motion artefact rejection and channel quality-checks the data were concatenated and mean-corrected. The data power spectra were then extracted using a fast Fourier transform. The LFOs were subdivided in 3 bands; very LFOs (VLFOs) 0.009-0.02 Hz, the resting state network (RSN) 0.02-0.1 Hz and the Mayer waves 0.1- 0.5 Hz. We then applied a Bonferroni-corrected paired t-test and a two-tailed t-test to compare the mean spectral power content of the left hemisphere versus right hemisphere channels for the different LFO bands.
Results: All infants with PAIS had a significantly lower mean spectral power in one or more of the LFO frequency bands over the injured hemisphere. Subject 1 exhibited lower mean power over the left hemisphere compared to the right in the VLFO band (p=0.005); subject 2 had a lower mean power over the left hemisphere compared to the right in the Mayer wave band (p=0.0056); subject 3 had a lower mean power over the left hemisphere compared to the right in all the bands (p=0.0037, p=0.0164, p=0.00018 respectively); and subject 4 had a lower mean power over the right hemisphere compared to the right in the RSN and Mayer wave bands (p=0.00098, p=0.0075 respectively). No significant inter-hemispheric differences were observed in the two control groups.
Conclusion: Our results suggest a link between the site of PAIS and the loss of certain LFOs, potentially secondary to altered sympathetic tone, altered vascular smooth muscle architecture or other neurogenic or metabolic factors. These findings require further investigation, but there appears to be potential for DOT in the early detection of PAIS.
Corresponding author: firstname.lastname@example.org
1. Obrig et al. Neuroimage 12(6), 2000.
2. White et al. Neuroimage 63(1) 2012.
3. Chalia et al. Neurophoton. 3(3), 031408 2016.