Purpose: Transforming intracranial pressure (ICP) into frequency domain commenced in the early 1980s, arriving at the conclusion that cerebrospinal dynamics were mapped by ICP spectral composition. Classical analysis tools were not suitable for handling intrinsic signal non-stationarity. To overcome inherent obstacles we introduce a novel approach based upon wavelets.
Methods: During routine diagnostic volume pressure testing epidural ICP was acquired in 118 patients with suspected cerebrospinal fluid circulatory disorders. Pressure was digitised and conditioned to separate low frequent signal components (<heart rate). ICP fluctuations were computed by subtraction of original and low frequent ICP constituents. Subsequently, multiresolution analysis was performed on fluctuations by discrete Haar wavelet transform and coefficients displayed in dyadic fashion (scalogram).
Results: Decomposition of ICP fluctuations led to typical patterns in the scalogram. Episodes of pathological wave activity and artificial ICP changes were topographically detectable in the time frequency plane.
Conclusions: The wavelet approach is a simple yet powerful signal processing method to estimate both static and dynamic properties of ICP in various clinical scenarios. It therefore outclasses classical spectral transforms that are limited to analysing real-world data. Haar wavelets are fast and robust. Their disadvantages seem not to counterbalance the advantages in this biomedical application.