gms | German Medical Science

Monitoring of cerebral haemodynamics is a crucial issue in the critical care of patients with severe head injury and/or subarachnoid hemorrhage. For this purpose, continuous monitoring includes measurement of the arterial blood (ABP) and intracranial pressure and cerebral blood flow velocity (CBFV). In the past, data analysis was restricted to a compromise between a high resolution in either time or frequency, e.g. FFT-analysis. In order to enhance monitoring of critical traumatic brain injury patients we applied the Wigner distribution, which allows a high temporal resolution of a frequency spectrum. Assuming a high pass model, efficiency of autoregulatory mechanisms was expected to decrease in frequency. Aim of the present study was to derive parameters that allow a classification of the status of cerebral autoregulation (CA).

We measured 14 subjects with either severe head injury or subarachnoid hemorrhage (age 38 ± 17 years) and 9 healthy volunteers (age 25 ± 5 years) for 20 minutes. Arterial blood pressure and intracranial pressure were recorded invasively as a part of standard monitoring. Cerebral blood flow velocity was determined by transcranial Doppler ultrasound in the middle cerebral artery in both hemispheres. After preprocessing of the data, the Wigner distribution was applied to the signals. By comparing the power distribution of ABP and CBFV in a low (0 – 0,07 Hz) and a high frequency band (0,2 – 0,4 Hz) we derived a time resolved expression of the power transfer. In the low frequency band, an efficient CA was expected to show a constantly lower power transfer from ABP to CBFV than in the high frequency band.

It could be shown that Mayer and B waves do not occur as stationary frequency components. Furthermore the new method allowed a continuous monitoring of cerebral autoregulation. Within a monitoring time of 20 min the power transfer from ABP to CBFV in normal subjects was for (19.2±1) min higher in the upper frequency band than in the low frequency band. In subjects with presumably impaired CA this was only the case for (12.4±2.2) min. This difference enabled us to differentiate significantly between subjects with impaired and preserved CA (t- Test: p < 0.01).

FFT-based processing can only be applied to stationary signals. The dynamic character of cerebral haemodynamics justifies the application of the Wigner distribution. The new method allows CA to be described as a function of time with high temporal resolution of 1 second. This offers completely new aspects in the analysis of cerebral haemodynamics.