IJCNMH ARCpublishing

Issue: Issue 1 (2014) – Supplement 1

Original Article

Convergent cross mapping: a promising technique for cerebral autoregulation estimation

Linda Heskamp, Aisha S.S. Meel-van den Abeelen, Joep Lagro, and Jurgen A.H.R. Claassen
Background: Cerebral autoregulation (CA) is the physiological mechanism that keeps the cerebral blood flow velocity (CBFV) relatively constant despite changes in arterial blood pressure (ABP). Currently, transfer function analysis (TFA) is widely used to assess CA non-invasively. TFA is based on the assumption that CA is a linear process, however, in reality CA is a non-linear process. This study explores the usability of convergent cross mapping (CCM) as a non-linear analysis technique to assess CA.

Methods: CCM determines causality between variables by investigating if historical values of a time-series X(t) can be used to predict the states of a time-series Y(t). The Pearson correlation is determined between the measured Y(t) and the predicted Y(t) and increases with increasing time-series length to converge to a plateau value. When used for CA, normal and impaired CA should be distinguishable by a different plateau value. With impaired CA, ABP will have a stronger influence on CBFV, and therefore the CBFV signal will contain more information on ABP. As a result, the correlation converges to a higher plateau value compared to normal CA. The CCM method was validated by comparing normal CA (normocapnia: breathing 0-2% CO2) with a model of impaired CA (hypercapnia: breathing 6-7% CO2).

Results: CCM correlation was higher (p=0.01) during hypercapnia (0.65 ± 0.16) compared to normocapnia (0.51 ± 0.18).

Conclusion: CCM is a promising technique for non-linear cerebral autoregulation estimation.

Keywords: Cerebral autoregulation, Convergent cross mapping, Non-linear analysis.

International Journal of Clinical Neurosciences and Mental Health 2014; 1(Suppl. 1):S20

Icon_pdf Download PDF