Signal analysis of dynamic magnetic resonance image acquisitions for the study of subtle blood-brain-barrier changes in small vessel disease

有证据表明，血脑屏障（BBB）的轻微破坏是包括脑小血管疾病和某些痴呆症在内的多种疾病的病理生理学组成部分。动态对比增强MRI（DCE-MRI）和示踪剂动力学建模用于评估BBB泄漏（Heye等人，2014）。然而，在渗漏轻微的疾病中，BBB渗漏的药代动力学模型有限，因为增强的幅度和速率较低，并且计算渗透性所需的微血管表面积未知（Heye等人，2016）。此外，扫描仪信号漂移、组织T1变化和伪影等因素可能会在估计的渗透性中引入系统误差，特别是在低渗透性时（Heye等人，2016）。这个问题使得不清楚信号增强的差异是否是由于微妙但关键的BBB异常。需要更好的方法将BBB泄漏的实际信号与“噪声”分离。最近，对比前后组织的纹理特征分析成为研究BBB破坏的潜在实用分析工具（Valdés Hernández等人，2017）。然而，尽管这种方法需要进一步发展，但它提供了一种潜在的稳健方法来区分BBB功能障碍的细微水平，以改善临床试验中的患者选择和分层，监测治疗并预测结果。使用纹理分析和其他信号处理方法重新评估患者的DCE-MRI数据，这些数据具有良好的认知、中风和小血管疾病特征，以区分BBB泄漏的严重程度2。将基于信号处理的最佳方法与通过传统方法获得的先前数据进行比较，并使用合成和临床数据3评估结果。提出一种实用的方法来分析临床试验中的微妙BBB泄漏。假设：我们假设结合多尺度主成分分析、去噪和从小波包分解信号子带中提取的高阶统计特征，将改善DCE-MRI对细微BBB泄漏的检测。方法：该项目将使用来自具有长期结局的良好特征患者（n=200）和正在进行的研究（n=200）的数据，这些研究具有常规BBB分析可用的DCE-MRI数据。待测试的高级信号处理方法将包括分析信号的功率谱（图1），寻求常见和疾病阶段特征空间模式之间的差异，并使用信号分解方法（例如经验模式分解，离散小波变换，小波包分解）来检查解剖学和病理学不同脑区的对比信号-时间轨迹。待测试的方法之一，精炼复合多尺度分散熵，是一种快速，强大的方法来量化信号复杂性（Azami H和Escudero J等人。2017），通过区分不同类型的动力学，该方法被证明有助于分析生理信号。参考文献- Heye，A. K.使用动态对比增强MRI评估血脑屏障破坏。系统综述。（2014）Neuroimage Clin; 6：262-274- Heye，A.K.等，Tracer kinetic modeling for DCE-MRI quantification of minute blood-brain barrier permeability.（2016）Neuroimage;一百二十五：446-455- Valdés Hernández et al.纹理分析在研究小血管疾病和血脑屏障完整性中的应用。（2017）前神经。https://doi.org/10.3389/fneur.2017.00327-Azami，H，Rostaghi，M，Abásolo，D & Escudero，J. Refined Composite Multiscale Dispersion Entropy and its Application to Biomedical Signals.（2017）IEEE生物医学工程学报（电子出版前打印）。DOI：10.1109/TBME.2017.2679136

SummaryThere is evidence that subtle breakdown of the blood-brain barrier (BBB) is a pathophysiological component of several diseases, including cerebral small vessel disease and some dementias. Dynamic contrast-enhanced MRI (DCE-MRI) and tracer kinetic modelling are used to assess BBB leakage (Heye et al. 2014). However, in diseases where leakage is subtle, pharmacokinetic models of the BBB leakage are limited since the magnitude and rate of enhancement are low and microvessel surface area, necessary to calculate permeability, is not known (Heye et al. 2016). Also, factors such as scanner signal drift, variations in tissue T1, and artefacts, can introduce systematic errors in estimated permeability, particularly at low permeability (Heye et al. 2016). This issue makes it unclear whether differences in signal enhancement are due to subtle but critical BBB abnormality or not. Better methods to separate the actual signal of BBB leakage from 'noise' are needed.The analysis of the textural features of the tissues pre and post contrast recently emerged as a potential, practical, analysis tool to study BBB disruption (Valdés Hernández et al. 2017). However, although this approach requires further development, it offers a potentially robust way to differentiate subtle levels of BBB dysfunction to improve patient selection and stratification in clinical trials, monitor treatment, and predict an outcome.Aims:1. Using Texture Analysis and other methods of signal processing to re-evaluate DCE-MRI data from patients well-characterised for cognition, stroke and small vessel disease, to differentiate severities of BBB leakage2. Compare the best signal-processing-based approach against previous data obtained by conventional methods, and evaluate the results using synthetic and clinical data3. Propose a practical approach to analyse subtle BBB leakage in clinical trials. Hypothesis:We hypothesise that combining multiscale principal component analysis, denoising, and higher order statistics features extracted from wavelet packet decomposition signal sub-bands, will improve detection of subtle BBB leakage with DCE-MRI. Method:The project will use data from well-characterised patients with long-term outcomes (n=200) and from ongoing studies (n=200 during the PhD) with DCE-MRI data in which conventional BBB analyses are available. The advanced signal processing methods to be tested will include analysis of the power spectrum of the signal (Figure 1), seeking differentiation between common and disease-stage-characteristic spatial patterns and using signal decomposition methods (e.g. empirical mode decomposition, discrete wavelet transform, wavelet packet decomposition) to examine the contrast signal-time trajectory in anatomically and pathologically different brain regions. One of the methods to be tested, Refined Composite Multiscale Dispersion Entropy, is a fast, powerful method to quantify signal complexity (Azami H and Escudero J et al. 2017), which proved useful to analyse physiological signals through distinguishing different types of dynamics. References- Heye, A. K. et al. Assessment of blood-brain barrier disruption using dynamic contrast-enhanced MRI. A systematic review. (2014) Neuroimage Clin; 6: 262-274- Heye, A.K. et al. Tracer kinetic modelling for DCE-MRI quantification of subtle blood-brain barrier permeability. (2016) Neuroimage; 125: 446-455- Valdés Hernández et al. Application of Texture Analysis to Study Small Vessel Disease and Blood-Brain Barrier Integrity. (2017) Front Neurol. https://doi.org/10.3389/fneur.2017.00327- Azami, H, Rostaghi, M, Abásolo, D & Escudero, J. Refined Composite Multiscale Dispersion Entropy and its Application to Biomedical Signals. (2017) IEEE Transactions on Biomedical Engineering (e-pub ahead of print). DOI:10.1109/TBME.2017.2679136