ICP waveform analysis

The third estimate of cerebrospinal compliance (denoted C_P1/P2 from the P1/P2 amplitude ratio) was based on the analysis of the ICP pulse waveform. Prior to the analysis, ICP, CBFV, and ABP (where available) signals were filtered using a low-pass filter with the cut-off frequency of 10 Hz. Individual pulses in the ICP signal were identified using a modified Scholkmann algorithm [4]; corresponding sections of CBFV and ABP signals were extracted based on pulse onset locations from ICP. For the purpose of peak annotation, each pulse was normalized and linearly detrended, and the three signals were aligned with regard to pulse onset in order to remove the phase shift resulting from distance between measurement sites.

A semi-automated algorithm based on the detection of the local maxima was used for peak identification. The algorithm incorporated information about the local maxima of ABP and CaBV (derived from the CBFV signal using Eq. (8)), taking into account observations from previous studies which showed that P1 is associated with the propagation of the pulse pressure wave through cerebral arteries and usually coincides with the systole of ABP, while P2 is derived from arterial blood volume load and coincides with the maximum of CaBV [8, 15]. In each pulse waveform, the local maxima of ICP corresponding to the position of maxima in ABP and CaBV were selected as candidates for P1 and P2, respectively. In the absence of the ABP signal, the first maximum of the CBFV signal was used instead to identify P1 candidates. Full detection results were reviewed and manually corrected in cases of insufficient detection accuracy, particularly in pathologically rounded waveforms. Pulses with distorted ICP waveform or unidentifiable P1 and P2 were excluded from further analyses. Illustrative examples of ICP pulse waveforms with peak annotations are presented in Fig. ​Fig.22.

The amplitude of peaks P1 and P2 in each pulse was calculated as the vertical distance to the preceding local minimum identified as pulse onset. Pulse-by-pulse P1/P2 amplitude ratio was then used as a compliance estimate.