Neonatal Brain Temperature Monitoring Based on Broadband Near-Infrared Spectroscopy.

Abstract

We present here the initial development of a novel algorithm based on broadband near-infrared spectroscopy (bNIRS) data to estimate the changes in brain temperature (BT) in neonates. We first explored the validity of the methodology on a simple numerical phantom and reported good agreements between the theoretical and retrieved values of BT and hemodynamic parameters changes, which are the parameters usually targeted by bNIRS. However, we noted an underestimation of the absolute values of temperature and haemoglobins' concentration changes when large variations of tissue saturation were induced, probably due to a crosstalk between the species in this specific case. We then tested this methodology on data acquired on 2 piglets during a protocol that induces seizures. We showed that despite a decrease in rectal temperature (RT) over time (-0.1048°C 1.5h after seizure induction, 95% CI: -0.1035 to -0.1061°C), BT was raising (0.3122°C 1.5h after seizure induction, 95% CI: 0.3207 to 0.3237°C). We also noted that the piglet displaying the largest decrease in RT also displays the highest increase in BT, which could be a marker of the severity of the seizure induced brain injury. These initial results are encouraging and show that having access to the changes in BT non-invasively could help to better understand the impact of BT on injury severity and to improve the current cooling methodologies in the neonatal neurocritical care following neonatal encephalopathy.