Development and application of signal analysis methods for preclinical and fundamental research in epilepsy

癫痫是一种以癫痫发作为特征的神经系统疾病，癫痫发作是突然的过度神经元放电，有时伴有意识丧失。它是一种常见的神经系统疾病，影响约1%的世界人口，约占总医疗费用的0.5%。虽然在了解和使用抗癫痫药物和手术治疗癫痫方面取得了进展，但尽管进行了所有治疗努力，仍有大约30%的癫痫患者继续遭受癫痫发作的破坏性影响。众所周知，癫痫发生（癫痫作为一种疾病发展的过程）和癫痫发生（个体癫痫发作的进展）的机制非常复杂和多样。与最终治疗特别相关的是，在分子和细胞水平上对疾病的详细了解与在世患者的临床成像数据之间弥合科学差距。这种桥梁主要依赖于局部神经元活动和血管反应之间的耦合，这是通过非侵入性成像技术观察到的。其中一种成像技术是近红外光谱（NIRS）。近红外光谱测量与神经活动相关的含氧和脱氧血红蛋白浓度的变化。与其他成像技术相比，它具有明显的优势，例如长期监控、便携性、安全性和低成本，使其成为某些应用中唯一的成像竞争者。在过去的几年里，我们小组在近红外光谱应用于癫痫的研究活动中发挥了主导作用。我们证明了近红外光谱足以检测各种癫痫发作，我们研究了癫痫尖峰，量化了它们的血流动力学反应的非线性。由于我们发现近红外光谱在检测癫痫发作方面是有效的，因此我们建议开发这种实时监测技术，单独使用或与脑电图（EEG）结合使用，用于患者的长期监测。这一努力最终可能在5到10年的时间里产生一种新的临床工具。该项目的一部分是开发和验证一个实时框架，用于用近红外光谱（和脑电图）检测癫痫发作。为了使这一努力取得成功，由于在癫痫患者中观察到的非常复杂的反应模式，反映在癫痫发作检测的假阳性率目前过高，无法进行强有力的临床应用，因此应在减少混淆方面取得进一步进展。我们假设动物模型可能是更好地理解这些模式的最佳策略，从而使人类漫射光学成像观察更加可靠。在小鼠中，我们将在用神经毒素诱导癫痫发作后进行先进的侵入性显微成像实验。他们稍后将扩展到更现实的人类癫痫模型，即中颞叶癫痫（MTLE）模型，注意到在我们的数据中，MTLE患者表现出特别隐式的近红外反应。利用光遗传学技术，在毫秒时间尺度上控制神经元尖峰，我们将测量在光遗传学刺激下神经元和血液对癫痫发作的反应，同时监测心率、呼吸频率和外周氧饱和度等关键生理变量。然后，我们将使用复杂的信号处理工具，包括生物物理模型来理解这些观察到的行为。这将有助于将动物的微观认识与人类的宏观观察联系起来。最终的结果将是一种临床工具，可以具体地改善住院患者的护理和生活，这些患者有复发性癫痫发作、中风和脑氧不饱和的风险。

Epilepsy is a neurological disorder characterized by epileptic seizures, which are sudden excessive neuronal discharges, sometimes accompanied by loss of consciousness. It is a common neurological disease, affecting approximately 1% of the world population and representing about 0.5% of total medical costs. While progress has been achieved in understanding and treating the epilepsies with anti-epileptic drugs and surgery, about 30% of epileptics continue to experience the devastating impact of seizures despite all efforts at treatment. It is well recognized that the mechanisms of epileptogenesis (the process of developing epilepsy as a disease) and of ictogenesis (the progression of individual seizures) are quite intricate and varied. Particularly relevant for eventual therapies is bridging the scientific gap between exquisitely detailed understanding of the disease at the molecular and cellular levels with clinical imaging data in living patients. Much of this bridge rests on the coupling between local neuronal activity and the vascular response that is observed with non-invasive imaging techniques.One such imaging technique is near-infrared spectroscopy (NIRS). NIRS measures changes in oxygenated and deoxygenated hemoglobin concentrations associated with neural activity. It offers distinct advantages over other imaging techniques, such as long-term monitoring, portability, safety, and low cost, making it the only imaging contender in some applications. Over the past few years, our group has taken a lead role in the research activity applying NIRS to epilepsy. We demonstrated that NIRS is adequate to detect various kinds of seizures and we studied epileptic spikes, quantifying the nonlinearity of their hemodynamic response. Since we found that NIRS was effective at detecting seizures, we propose here to develop this real-time monitoring technology, by itself or in combination with electroencephalography (EEG), for long-term monitoring of patients. This effort could eventually lead to a new clinical tool over a 5 to 10 year horizon. One part of this project is the development and validation of a real-time framework for detecting seizures with NIRS (and EEG). For this effort to succeed, further progress should be made for reducing the confounding aspects due to the very complex response patterns observed in epileptic patients, reflected in false positive rates for seizure detection that are too high at present for a robust clinical application. We hypothesize that an animal model may be the best strategy to better understand these patterns so as to make the diffuse optical imaging observations in human more reliable. In mice, we will perform advanced and invasive microscopic imaging experiments after inducing epileptic seizures with a neurotoxin. They will be extended later to a more realistic model of human epilepsy, a mesio-temporal lobe epilepsy (MTLE) model, noting that in our data, MTLE patients exhibit especially cryptic NIRS responses. Using optogenetic technology, to control neuronal spiking at the millisecond time scales, we will measure the neuronal and the blood responses to seizures when perturbed by optogenetic stimulation, while monitoring key physiological variables such as heart rate, respiration rate and peripheral oxygen saturation. We will then use sophisticated signal processing tools including biophysical models to understand these observed behaviors. This will help bridge the microscopic understandings in animals with the macroscopic observations in humans. The final result will be a clinical tool that could concretely improve care and the lives of patients admitted to the hospital at risk for recurrent seizures, stroke and brain oxygen desaturation.