Cerebral tissue oxygen saturation calculated using low frequency haemoglobin oscillations measured by near infrared spectroscopy in adult ventilated patients.

Oxy- (HbO2) and deoxy- (HHb) haemoglobin signals measured by near infrared (NIR) spectroscopy over the human frontal lobes frequently contain respiratory and low frequency oscillations (LFOs). It has been suggested previously that venous oxygen saturation (SvO2) can be calculated from these respiratory oscillations. In this paper, we investigated the use of a Fourier transform based algorithm to calculate an oxygen saturation measure known as S(osc)O2 which may be a close estimate of the underlying SvO2. S(osc)O2 was calculated using three different frequency ranges, (1) respiratory oscillations only, (2) LFOs only, and (3) both respiratory oscillations and LFOs. At each frequency range S(osc)O2 was calculated using either (1) the modified Beer-Lambert law (MBL) or (2) spatially resolved spectroscopy (SRS). In total six different measurements of S(osc)O2 were investigated here. Experiments were performed in six adult ventilated patients with traumatic brain injury. The patients' inspired oxygen fraction (FiO2) was raised in two hyperoxic phases. The calculated S(osc)O2 values were compared with other cerebral oxygenation measures including an intraparenchymal catheter based brain tissue oxygen tension (PbrO2) and the NIR based tissue oxygenation index (TOI). It was found that the S(osc)O2 calculated using the combined respiratory and LFO frequency range and the SRS method resulted in the highest detection rates of hyperoxic changes. This measure of S(osc)O2 may provide a viable, continuous, non invasive, bedside measure of cerebral venous oxygen saturation.