Chapter 7 frequency trend, of both signal needs to be removed. This was either performed by mean removal or using a smoothness priors detrending method 35. Spectral estimates can be calculated either through epoch averaging using the Welch method 39 or by spectral smoothing 37. For both several parameter settings need to be chosen, such as window length, overlap percentage, tapering. Combining detrending and spectral estimate methods four different signal processing combinations have been compared. For dCA parameters evaluated with the four different combinations in chapter 2 no differences have been found. This could lead to the conclusion that it makes no difference which parameter settings are chosen for TFA. However, this was only investigated with data using either spontaneous blood pressure variations or paced breathing induced blood pressure variations. Furthermore, the four parameter combinations compared are not all possible options. Currently, initiated by the Cerebral Autoregulation Research Network (CARNet, www.car-net.org), a multicenter study (11 participating centers including our hospital) is performed to compare different TFA methods for quantifying dCA. Each center will analyze the same data set using their own standard way of analyzing dCA recordings. Results will be compared between centers to evaluate the influence of different TFA parameter settings. The data set used in this study only consists of recordings using spontaneous blood pressure variations. When dCA is being challenged more by higher amplitude ABP variations e.g. by squat-stand maneuvers possibly other effects of different TFA parameter settings could arise. Hopefully, this study and the cooperation in this CARNet research group will contribute to standardization of TFA in dCA. In single input-single output TFA, coherence is an indication for the fraction of output variation that can linearly be explained by the input variation. For a perfect linear input-output relation coherence equals one. Coherence can be lower than one for several reasons: 1) if there is noise involved, resulting in input and/or output variation that is not explained by the linear transfer function, 2) if other variation from an additional input contributes to the output; multiple coherence 23 can be used to investigate this, 3) if the transfer function is non-linear; non-linear techniques may be used 12, 15-18, 21. Giller, who was the first to investigate frequency-dependent behavior of cerebral autoregulation, suggested coherence as a measure of dCA 10. With intact CA, due to changes in CVR, changes in ABP will be attenuated in CBFV, resulting in low coherence, whereas at the other extreme, if CA is impaired, CBFV will tend to follow the changes in ABP leading to high values of 126

proefschrift gommer inhoud met kaft.indd

To see the actual publication please follow the link above