Research On Dynamic Assessment Of Cerebral Autoregulation

Cerebral Autoregulation (CA) can be defined as the physiological regulatory mechanism that maintains a relatively constant cerebral blood flow (CBF) despite wide ranging changes in arterial blood pressure (ABP). Impaired autoregulation, which may lead to ischemia and hyperaemia in the brain, is associated with pathophysiological conditions including intracranial tumours, head injury, hypertension and space-occupying lesions in the brain. Therefore, researching CA is important to guide the treatments of patients suffering from such conditions. Linear and nonlinear model are used in researching on CA, linear models are applied to different physiological and pathological, and autoregulation index(ARI) and frequency-domain parameters (Phase-lead) caculated from linear models have been shown to sensitive to pathophysiological changes in patients with carotid artery disease, storke, sever head injury and other conditions. In this paper, based on continuous ABP and CBF acquired by measurements with high temporal resolution, such as transcranial Doppler ultrasonography (TCD), by utilizing Recursive Least Squares method (RLS), a six-order Finite Impluse Response (FIR) model is builted. This Model and its parameters are also tested and verified in the paper. ARI and Phase-lead are the indicators of evaluating CA.Main works included in this paper:(1) An interface about displaying data and marking abnormal data is designed, manual and automatic abnormal signal marking process is realized, completing the experimental data screening work. Due to certain reasons about our experimental facilities, the 62 groups of data we collected contain some abnormal signals. No cerebral autoregulation information is contained in these abnormal signals; they need to be screened out. Considering the clinical practicability, An automatic marking method for detecting abnormal signal in ABP and CBF waveforms is designed, with the capability of efficiently eliminating abnormal signal by comparing the difference between normal and abnormal signals within frequency domain.(2) Evaluation indicators analysis, caculation about ARI and Phase-lead. When caculate ARI, the frequency about Tiecks's model should be fixed at 10 Hz, samples of step response among 1.5~3.5s are selected for verification purpose. Characteristics about Phase-lead include the realization of transfer function and determination about the frequency range when caculating Phase-lead.(3) Cerebral autoregulation model building and dynamic model parameters identification. Model identification tool box is used to help identify the structure of model,, the six-order Finite Impluse Response (FIR) model is chosen as the method for identifying the model after comparing and analyzing Infinite Impluse Response (IIR) model with Tiecks model and IIR model with FIR model. 15 groups Blood pressure and blood flow data under spontaneous fluctuations and 20 groups of that under the condition of CO2 induced sudden changes of autoregulation are tracked by RLS. Research shows that RLS method is effective when used to track data under spontaneous fluctuation, but it performs not so effective when the autoregulation changes suddenly.