Application of Machine Vision to Determine the Influence of Sleep States and Social Interactions on Vulnerability to Drug Addiction

项目总结/摘要 该项目的长期目标是开发一种技术，使成瘾研究人员能够测量 在大规模基因组实验中，社会和睡眠行为对药物摄入行为的影响 发现调节成瘾脆弱性的基因。为了实现这一目标，我们将开发并公开 这是一项高度创新的技术，首次允许研究人员连续监测 复杂的睡眠和社会行为的小鼠群体，使用低成本和易于缩放的视频设备。而 药物成瘾行为与这些行为有着良好的双向关系，遗传 病因尚未完全阐明。这一关键差距主要是由于技术壁垒， 对大量动物的特定睡眠状态和社会行为进行可靠的表型分析。目前 用于评估不同睡眠状态的可用方法被设计成用于隔离的动物，并且不 测量一组中的多个个体。尽管社会环境对成瘾有影响 由于人类的脆弱性，人们主要通过改变居住环境和测试隔离的动物来进行研究。我们 将利用人工智能技术开发基于神经网络的视频机器视觉方法 这些行为的分析，旨在用于一个与行为学相关的群体环境，在很长一段时间内， 时间这种非侵入性的方法将是行为表型分析的一个重大进步， 高通量遗传学研究的必要性，以优化小鼠作为成瘾模型。我们的第一个具体 目的是开发一种机器视觉方法来测量快速眼动（REM）睡眠，非REM（NREM） 所有个体的睡眠和清醒行为，用于单个或群体圈养条件。我们将训练 并使用EEG和EMG记录来验证我们的机器视觉网络。为了展示实用性并评估 我们的方法的性能，我们将比较两种不同的小鼠系（对照小鼠C57 B1/6 J和 基因改变的菌株，B6.129P2-Nos 2 tm 1 Lau/J），其已知在这些睡眠参数上不同。我们将 比较这两种小鼠品系在分组和单独饲养中的睡眠。第二个具体目标是 扩展我们的方法，使评估组的社会动态，并记录所有的社会和积极的行为 持续了好几天这将使我们能够测试社交和睡眠行为的双向影响， 吸食自己服用的甲基苯丙胺我们将评估初始社会地位和睡眠的影响 药物消费的质量，并衡量社会互动和睡眠模式如何改变后， 药物是可用的。成功完成这些目标将产生一种经过验证的技术， 提供了用于大规模遗传研究的小鼠睡眠和社会行为的详细测量结果， 从而大大提高了研究人员识别与成瘾易感性相关的基因的能力。

PROJECT SUMMARY/ABSTRACT The long-term goal of this project is to develop a technology that will allow addiction researchers to measure the influence of social and sleep behaviors on drug intake behavior in large-scale genomic experiments to discover genes that regulate vulnerability to addiction. To accomplish this we will develop and make publicly available a highly innovative technology that, for the first time, will allow researchers to continuously monitor complex sleep and social behaviors in groups of mice, using low-cost and easily scaled video equipment. While drug-addiction behaviors have well-established bidirectional relationships with these behaviors, the genetic etiologies have not been fully elucidated. This critical gap is primarily due to technological barriers that prevent reliable phenotyping of large numbers of animals for specific sleep states and social behaviors. Currently available methods for assessing different sleep states are designed to be used with isolated animals and do not measure multiple individuals in a group. Although social context is known to be an influence on addiction vulnerability, it has been largely studied by changing the housing environment and testing isolated animals. We will exploit techniques of artificial intelligence to develop a neural network-based machine-vision method of video analysis of these behaviors, designed to be used in an ethologically-relevant group setting, over long periods of time. This non-invasive method will be a significant advance in behavioral phenotyping, fulfilling the demands of the high-throughput genetic studies necessary for optimizing the mouse as model of addiction. Our first specific aim is to develop a machine-vision method to measure rapid eye movement (REM) sleep, nonREM (NREM) sleep and waking behaviors of all individuals, for use in either single or group housed conditions. We will train and validate our machine-vision networks using EEG and EMG recordings. To demonstrate the utility and assess the performance of our method we will compare two different mouse lines (the control mouse C57Bl/6J and a genetically altered strain, B6.129P2-Nos2 tm1Lau/J) that are known to differ in these sleep parameters. We will compare the sleep of these two mouse strains in both group and single housing. Our second specific aim is to extend our method to enable assessment of group social dynamics and to record all social and active behaviors over multiple days. This will allow us to test the bidirectional effects of social and sleep behaviors with the consumption of self-administered methamphetamine. We will assess the effect of initial social status and sleep quality on drug consumption and also measure how the social interactions and sleep patterns are changed after the drug is available. Successful completion of these aims will yield a validated technology with the capacity to provide detailed measurements of sleep and social behaviors in mice for use in large scale genetic studies, thereby significantly enhancing researchers' ability to identify genes associated with addiction susceptibility.