Estimation of the Retinotopic Map of an Awake Mouse Brain Based upon Intrinsic Optical Signal Imaging Considering the Ocular Position and Variation in Pupil Diameter
Estimation of the Retinotopic Map of an Awake Mouse Brain Based upon Intrinsic Optical Signal Imaging Considering the Ocular Position and Variation in Pupil Diameter
复制标题
考虑眼位置和瞳孔直径变化的基于本征光信号成像的清醒小鼠大脑视网膜专题图的估计
DOI：
10.11239/jsmbe.56.157
复制
发表时间：
2018
影响因子：
--
通讯作者：
片山 統裕
中科院分区：
文献类型：
--
作者：
外川 龍之介;奥畑 大悟;吉田 侑冬;中尾 光之;片山 統裕
我们发展了一种新的方法，从清醒小鼠视觉刺激诱导的内在光信号（IOS）中估计初级视皮层的精细视网膜定位图。与使用麻醉的方法不同，为了减轻动物的负担，缩短实验时间是一个重要的要求。在清醒状态期间，存在眼球运动、瞳孔直径波动和大脑背景活动。眼球运动的发生使视网膜图像模糊。在传统方法中，在同步平均方法中排除这种情况下的数据是必要的。为了解决这些问题，我们针对瞳孔直径与IOS的全局信号（GS）之间的强相关性，提出了一种在预处理中去除IOS中GS的方法。这一过程在一次试验中提高了视觉反应的SN比。我们假设来自皮层感兴趣区（ROI）的响应由视网膜图像与表示从视网膜到皮层的投影的感受野函数的乘积和来描述。在该模型中，与同步平均方法不同，眼睛运动的影响可以通过移动视网膜图像来表示。因此，所有的响应数据都可以用于估计参数，而与刺激位置或眼睛位置无关。另外，在该方法中，空间分辨率不取决于刺激点的空间分辨率。感受野函数的参数可以用非线性最小二乘法估计。将该方法应用于真实的数据，得到了比传统方法高得多的空间分辨率。有趣的是，类似于高级大脑区域的结构，如次级视觉皮层，以前只使用侵入性方法观察到，如钙成像和电生理方法与电极插入在小鼠大脑中，也被可视化。这些结果表明，所提出的方法具有高的空间分辨率的实用性。
We have developeda novel methodto estimate the fine retinotopic map of the primary visual cortex from the intrinsic optical signal (IOS) induced by visual stimulation in an awake mouse. Unlike methods employing anesthesia, in order to reduce the burden on the animal, shortening the experimental time is an important requirement. During the awake state, eye movement, pupil diameter fluctuations, and brain background activity are present. Occurrence of eye movement blurs the retinal image. Excluding data under such circumstances in the synchronous average method is essential in conventional methods. In order to solve these problems, we focusedon the strong correlation between the pupil diameter andthe global signal (GS) of IOS andintroduced a process to remove GS from IOS in preprocessing. This process improvedthe SN ratio of visual response in a single trial. We assumedthat the response from the region of interest (ROI) of the cortex is described by the product sum of the retinal image andthe receptive fieldfunction expressing the projection from the retina to the cortex. In this model, unlike the synchronous average method, the influence of eye movement can be expressed by shifting the retinal image. Therefore, all the response data can be used to estimate parameters, irrespective of the stimulation location or eye position. Additionally, in this method, the spatial resolution does not depend on the spatial resolution of the stimulation spot. The parameters of the receptive fieldfunction can be estimatedusing the nonlinear least squares method. By applying this method to real data, we obtained a retinotopic map with much higher spatial resolution than that obtainedby conventional methods. Interestingly, structures similar to higher brain regions such as secondary visual cortex, which were previously observed only using invasive methods such as calcium imaging andelectrophysiological methodwith electrode insertion in the mouse brain, were also visualized. These results demonstrate the usefulness of the proposed method with high spatial resolution.