The Effect of Data Length on the Assessment of Dynamic Cerebral Autoregulation with Transfer Function Analysis in Neurological ICU Patients.

BACKGROUND: Cerebral autoregulation plays an important role in safeguarding adequate cerebral perfusion and reducing the risk of secondary brain injury, which is highly important for patients in the neurological intensive care unit (neuro-ICU). Although the consensus white paper suggests that a minimum of 5 min of data are needed for assessing dynamic cerebral autoregulation with transfer function analysis (TFA), it remains unknown if the length of these data is valid for patients in the neuro-ICU, of whom are notably different than the general populations. We aimed to investigate the effect of data length using transcranial Doppler ultrasound combined with invasive blood pressure measurement for the assessment of dynamic cerebral autoregulation in patients in the neuro-ICU. METHODS: Twenty patients with various clinical conditions (severe acute encephalitis, ischemic stroke, subarachnoid hemorrhage, brain injury, cerebrovascular intervention operation, cerebral hemorrhage, intracranial space-occupying lesion, and toxic encephalopathy) were recruited for this study. Continuous invasive blood pressure, with a pressure catheter placed at the radial artery, and bilateral continuous cerebral blood flow velocity with transcranial Doppler ultrasound were simultaneously recorded for a length of 10 min for each patient. TFA was applied to derive phase shift, gain, and coherence function at all frequency bands from the first 2, 3, 4, 5, 6, 7, 8, 9, and 10 min of the 10-min recordings in each patient on both hemispheres. The variability in the autoregulatory parameters in each hemisphere was investigated by repeated measures analysis of variance. RESULTS: Forty-one recordings (82 hemispheres) were included in the study. According to the critical values of coherence provided by the Cerebral Autoregulation Research Network white paper, acceptable rates for the data were 100% with a length ≥ 7 min. The final analysis included 68 hemispheres. The effects of data length on trends in phase shift in the very low frequency (VLF) band (F1.801,120.669 = 6.321, P = 0.003), in the LF band (F1.274,85.343 = 4.290, P = 0.032), and in the HF band (F1.391,93.189 = 3.868, P = 0.039) were significant for 3-7 min, for 4-7 min, and for 5-8 min, respectively. Effects were also significant on the gain in the VLF band (F1.927,129.134 = 3.215, P = 0.045) for 2-8 min and on the coherence function in all frequency bands (VLF F2.846,190.671 = 90.247, P < 0.001, LF F2.515,168.492 = 55.770, P < 0.001, HF F2.411, 161.542 = 33.833, P < 0.001) for 2-10 min. CONCLUSIONS: Considering the acceptable rates for the data and the variation in the TFA variables (phase shift and gain), we recommend recording data for a minimum length of 7 min for TFA in patients in the neuro-ICU.

The consistency of invasive and non-invasive arterial blood pressure for the assessment of dynamic cerebral autoregulation in NICU patients.

Background: Studies of the clinical application of dynamic cerebral autoregulation show considerable variations, and differences in blood pressure devices may be one of the reasons for this variation. Few studies have examined the consistency of invasive and non-invasive arterial blood pressure for evaluating cerebral autoregulation. We attempted to investigate the agreement between invasive and non-invasive blood pressure methods in the assessment of dynamic cerebral autoregulation with transfer function analysis. Methods: Continuous cerebral blood flow velocity and continuous invasive and non-invasive arterial blood pressure were simultaneously recorded for 15 min. Transfer function analysis was applied to derive the phase shift, gain and coherence function at all frequency bands from the first 5, 10, and 15 min of the 15-min recordings. The consistency was assessed with Bland-Altman analysis and intraclass correlation coefficient. Results: The consistency of invasive and noninvasive blood pressure methods for the assessment of dynamic cerebral autoregulation was poor at 5 min, slightly improved at 10 min, and good at 15 min. The values of the phase shift at the low-frequency band measured by the non-invasive device were higher than those measured with invasive equipment. The coherence function values measured by the invasive technique were higher than the values derived from the non-invasive method. Conclusion: Both invasive and non-invasive arterial blood pressure methods have good agreement in evaluating dynamic cerebral autoregulation when the recording duration reaches 15 min. The phase shift values measured with non-invasive techniques are higher than those measured with invasive devices. We recommend selecting the most appropriate blood pressure device to measure cerebral autoregulation based on the disease, purpose, and design.

A Data-Driven Approach to Transfer Function Analysis for Superior Discriminative Power: Optimized Assessment of Dynamic Cerebral Autoregulation.

Transfer function analysis (TFA) is extensively used to assess human physiological functions. However, extracting parameters from TFA is not usually optimized for detecting impaired function. In this study, we propose to use data-driven approaches to improve the performance of TFA in assessing blood flow control in the brain (dynamic cerebral autoregulation, dCA). Data were collected from two distinct groups of subjects deemed to have normal and impaired dCA. Continuous arterial blood pressure (ABP) and cerebral blood flow velocity (CBFV) were simultaneously recorded for approximately 10 mins in 82 subjects (including 41 healthy controls) to give 328 labeled samples of the TFA variables. The recordings were further divided into 4,294 short data segments to generate 17,176 unlabeled samples of the TFA variables. We optimized TFA post-processing with a generic semi-supervised learning strategy and a novel semi-supervised stacked ensemble learning (SSEL) strategy for classification into normal and impaired dCA. The generic strategy led to a performance with no significant difference to that of the conventional dCA analysis methods, whereas the proposed new strategy boosted the performance of TFA to an accuracy of 93.3%. To our knowledge, this is the best dCA discrimination performance obtained to date and the first attempt at optimizing TFA through machine learning techniques. Equivalent methods can potentially also be applied to assessing a wide spectrum of other human physiological functions.

Assessment of dynamic cerebral autoregulation in patients with basilar artery stenosis.

BACKGROUND AND AIMS: Previous studies have shown impaired cerebral autoregulation (CA) in carotid and middle cerebral artery (MCA) stenosis/occlusion. Little is known about CA in patients with basilar artery (BA) stenosis. We therefore investigated dynamic CA patterns in BA stenosis using transfer function analysis (TFA). METHODS: We measured spontaneous oscillations of blood flow velocity (CBFV) in the right posterior cerebral artery (PCA), and left MCA and mean arterial pressure (ABP) continuously in 25 patients with BA stenosis (moderate n=16 with 50-69% occlusion and severe n=9 with ≥ 70% occlusion) and 22 healthy volunteers in supine position during 6 circles per minute deep breath. Analysis was based on the 'black-box' model of transfer function deriving phase and gain in both PCA and MCA. RESULTS: Though changes of phase shift and gain between the patients and healthy controls were observed in MCA, the differences are however not significant. Phase shift in PCA was significantly decreased in severe stenosis when comparing with healthy controls and moderate stenosis (4.2 ± 34.2° VS 41.1 ± 40.4°, 4.2 ± 34.2° VS 34.2 ± 27.2°, both p<0.05), whilst the gain in PCA is increased for moderate BA stenosis and decreased for severe BA stenosis. Furthermore, we found that phase shift were almost abolished in patients with ischemic stroke who developed unfavorable clinical outcome (mRs>2) on the 90 days after stroke onset. CONCLUSION: Dynamic CA in PCA reduces in patients with severe BA stenosis and those with ischemic stroke who present poor outcome in 90 days after stroke onset. Phase shift might be a sensitive index prompting impaired CA in posterior circulation.