Uma banca de QUALIFICAÇÃO de MESTRADO foi cadastrada pelo programa.
DATA : 20/06/2023
HORA: 10:00
LOCAL: Sala 107 Bloco Zeta Campus São Bernardo do Campo

Characterization of cerebral autoregulation in patients with stroke and hypertension


Introduction: The human brain relies on a consistent supply of oxygen and energy substrates through the cerebral blood flow (CBF). The occurrence of ischemic stroke can severely impact this supply by altering brain perfusion. Cerebral autoregulation (CA) maintains a constant cerebral blood flow (CBF) during changes in arterial blood pressure (ABP). However, if compromised, CBF becomes dependent on ABP resulting in critical conditions for the brain. Previous studies demonstrated that CA are compromised with increasing occurrence of stroke. Furthermore, Systemic arterial hypertension (SAH) is the major risk factor for stroke and is one the leading causes of morbidity and mortality worldwide. SAH may lead to structural and functional changes in the arteriolar wall resulting in an impairment of CBF regulation as well. The evaluation of CA is commonly done by mathematical approaches that use the mean values of ABP and CBF collected in the same period of time. Though, currently no gold standard test exits, many different methods are now available for CA assessment. However, the understanding of relationships among CA indices remains limited.

Objectives: This project aims to investigate the relationship between CA assessment methods and clinical data from patients with acute ischemic stroke and SAH, in order to produce a multidimensional analysis that relates these dimensions (CA indices) to physiological changes resulting from stroke and SAH. These data will then be compared to findings of a control group.   More specifically, the objectives are: apply four distinct CA assessment methods to a sample space consisting of patients with ischemic stroke, SAH and control subjects. Analyze the difference in CA results between the control group and the groups with different levels of stroke and SAH (mild, moderate and severe). Evaluate the feasibility of employing supervised classifying techniques to automatically predict stroke outcomes, based on CA methods.

Methods: For the study with ischemic stroke patients 50 control and 60 stroke subjects were enrolled. For the study with SAH patients, 30 SAH subjects were enrolled. For each subject, bilateral cerebral blood flow velocity (CBFv) and non-invasive arterial blood pressure (ABP) were collected (5 min, Fs: 100 Hz), low-pass filtered (cutoff:20 Hz, Butterworth, 3rd order), beat-to-beat mean value interval were extracted and the surrogated signal upsampled (5 Hz, cubic spline interpolation).

Cerebral Autoregulation analysis: CA analysis involves four CA assessment methods, namely Transfer Function analysis (TFA), Autoregulation index (ARI), ARI with Autoregressive moving average model (ARMA-ARI) and non-invasive Mean flow index (nMx). TFA is estimated through the relation between auto-spectrum of ABP and the cross-spectrum of ABP with CBFv, based on the discrete Fourier transform combined with Welch’s method applied to ABP and CBFv (Hanning window, 100 s and 50% overlap). Then, gain, phase and coherence are calculated in three different frequency bands, 0.02-0.07 Hz (very low Frequency, VLF), 0.07-0.2 Hz (low frequency, LF) and 0.2-0.5 Hz (high frequency, HF). ARI consists of 10 levels from damaged (ARI = 0) to intact (ARI = 9) CA. A second-order differential equation simulates 10 sets of CBFv responses to an ideal step change in BP. By comparing the recorded CBFv with the 10 simulated velocities, the ARI is assigned to each subject for both hemispheres by using the best least squares fit. ARMA-ARI is an alternative calculation of ARI, it was obtained by first modelling the dynamic relationship between ABP and CBFv with an autoregressive-moving average model (ARMA). The model orders AR and MA was defined as 2 and 3 respectively. To estimate these coefficients for each data segment, the least-square method is employed. nMx method consists of taking a sequence of consecutive samples of two time series and calculate Pearson's correlation coefficient over the interval.  Its value ranges from -1 (representing perfect negative correlation) to 0 (indicating no correlation) to 1 (representing perfect positive correlation). Therefore, nMx coefficient is calculated through Pearson's correlation between the mean values of ABP and CBFv signals, considering 300 seconds of recording, and windows nMx of 60 seconds.

Classification: The classification of stroke outcomes model is based on comparing the normative CA findings obtained by NIHSS index with CA methods (TFA, ARI, ARMA-ARI and nMx). Firstly, patients were ranked by a stroke scale (NIHSS: mild ≤4, moderate 5-15, and severe ≥16) by clinical specialists. Then, K-nearest neighbors’ method was applied on the data (i.e VLF/LF and HF of each subject vs. stroke scale) randomly divided into 75% training and 25% testing. The model's results were evaluated through accuracy (ACC) and ROC area under the curve (AUC).

Results: Stroke groups presented higher mRS, ABP and heart rate. Results of CA metrics showed that gain (VLF and LF), phase (HF), ARI and nMx presented significant differences between the stroke and control groups, being most of the differences between control and severe groups. Coherence presented values higher than 0.5 for all frequency ranges, indicating that the estimation of gain and phase parameters were reliable. In the SAH study, mean values between mild and severe groups presented no statistical difference for all CA methods. Initial classification analysis showed that ACC from gain (68% and 62% for VLF and LF, respectively); ARI-ARMA 68%, and phase 62% (LF) did perform better. Although most of the methods presented AUC close to the minimum acceptable (50%).

Conclusion: The results presented in this study add to existing evidence that cerebral hemodynamics are negatively affected in stroke patients, with increase stroke severity leading to greater CA impairment. Besides that, stroke severity can also be associated negatively with mRS outcome, indicating poor functioning recovery in patient with CA impairment. Nevertheless, SAH results demonstrated that sustained arterial hypertension does not alter CA within the ABP limits studied. Leading to the conclusion that, despite being a major risk factor for stroke, SAH is not expressed by CA. In the classification method, a large percentage of false outcomes still present in all four CA assessment methods.  Currently, other machine learning methods is under analysis, such as Support vector machine and artificial neural networks.  Outcomes will be present by the presentation, to highlight more accurate and reliable diagnostic for stroke patients

Presidente - Interno ao Programa - 2188954 - ERICK DARIO LEON BUENO DE CAMARGO
Membro Titular - Examinador(a) Externo ao Programa - 1188948 - JOAO LAMEU DA SILVA JUNIOR
Membro Titular - Examinador(a) Externo à Instituição - RICARDO DE CARVALHO NOGUEIRA
Membro Suplente - Examinador(a) Interno ao Programa - 2352005 - RENATO NAVILLE WATANABE
Membro Suplente - Examinador(a) Externo à Instituição - MATHEUS CARDOSO MORAES - UNIFESP
Membro Suplente - Examinador(a) Externo à Instituição - ALESSANDRO PEREIRA DA SILVA - UMC
Notícia cadastrada em: 15/05/2023 14:26
SIGAA | UFABC - Núcleo de Tecnologia da Informação - ||||| | Copyright © 2006-2024 - UFRN - sigaa-1.ufabc.int.br.sigaa-1-prod