Reconstruction of natural images from responses of primate retinal ganglion cells.
Reconstruction of natural images from responses of primate retinal ganglion cells.
复制标题
从灵长类视网膜神经节细胞的反应重建自然图像。
DOI：
10.7554/elife.58516
复制
发表时间：
2020-11-04
期刊：
影响因子：
7.7
通讯作者：
Chichilnisky EJ
中科院分区：
文献类型：
--
作者：
Brackbill N;Rhoades C;Kling A;Shah NP;Sher A;Litke AM;Chichilnisky EJ
视网膜神经节细胞（RGC）传达的视觉信息通常由其空间感受野概括，但原则上也取决于其他RGC的反应和自然图像统计。这种可能性进行了探讨，线性重建的自然图像的响应的四个数字占主导地位的猕猴RGC类型。视网膜重建高度一致。每个RGC的最佳重建滤波器-其视觉信息-反映了自然的图像统计数据，并且只有当附近的相同类型的细胞被包括在内时才与感受野相似。ON和OFF细胞主要传达独立的、互补的表征，而阳伞细胞和侏儒细胞传达不同的特征。相关活动和非线性有统计学意义，但对重建的影响较小。模拟重建，使用线性-非线性级联模型的RGC光响应，将测量的空间特性和非线性，产生了类似的结果。时空重建表现出类似的空间特性，这表明结果是相关的自然视觉。视觉始于视网膜，视网膜是眼睛后部的一层组织。被称为视杆细胞和视锥细胞的光敏细胞吸收入射光并将其转换为电信号。它们将这些信号传递给称为视网膜神经节细胞（RGC）的神经元，这些神经元将它们转化为称为尖峰的电信号。来自RGC的尖峰信号然后沿着视神经到达大脑。它们是大脑接收视觉信息的唯一来源。从这些信息中，大脑构建了我们的整个视觉世界。灵长类动物的视网膜包含大约20种类型的RGC。每一个编码一个不同的视觉特征，如一定大小的亮点的存在，或有关纹理和运动的信息。但是，每个RGC向大脑发送的确切信息，以及大脑如何使用这些信息，目前尚不清楚。Brackbill等人通过测量和分析猕猴视网膜的电活动来回答这些问题。研究猕猴视网膜很重要，因为灵长类动物的视觉系统在几个方面与其他物种不同。这些包括视网膜中存在的RGC的数量和类型。这些灵长类动物在高分辨率中央视觉和三色色觉方面也与人类相似。Brackbill等人使用电极阵列同时监测数百个RGC，记录了猕猴视网膜对现实生活中的风景，物体，动物或人的反应。基于这些记录，再加上关于RGC反应的现有知识，Brackbill等人然后试图仅使用记录的电活动来重建原始图像。所得到的重建在所有测试的视网膜中是相似的。此外，它们与原始图像惊人地相似。这些结果使人们有可能理解每个细胞的光响应特性如何代表大脑可以使用的视觉信息。了解猕猴视网膜在自然条件下如何工作对于解码我们自己的视网膜如何处理和传递信息至关重要。更好地了解大脑如何使用这种输入来生成图像，最终可以设计人工视网膜来恢复某些形式的失明患者的视力。
The visual message conveyed by a retinal ganglion cell (RGC) is often summarized by its spatial receptive field, but in principle also depends on the responses of other RGCs and natural image statistics. This possibility was explored by linear reconstruction of natural images from responses of the four numerically-dominant macaque RGC types. Reconstructions were highly consistent across retinas. The optimal reconstruction filter for each RGC – its visual message – reflected natural image statistics, and resembled the receptive field only when nearby, same-type cells were included. ON and OFF cells conveyed largely independent, complementary representations, and parasol and midget cells conveyed distinct features. Correlated activity and nonlinearities had statistically significant but minor effects on reconstruction. Simulated reconstructions, using linear-nonlinear cascade models of RGC light responses that incorporated measured spatial properties and nonlinearities, produced similar results. Spatiotemporal reconstructions exhibited similar spatial properties, suggesting that the results are relevant for natural vision. Vision begins in the retina, the layer of tissue that lines the back of the eye. Light-sensitive cells called rods and cones absorb incoming light and convert it into electrical signals. They pass these signals to neurons called retinal ganglion cells (RGCs), which convert them into electrical signals called spikes. Spikes from RGCs then travel along the optic nerve to the brain. They are the only source of visual information that the brain receives. From this information, the brain constructs our entire visual world. The primate retina contains roughly 20 types of RGCs. Each encodes a different visual feature, such as the presence of bright spots of a certain size, or information about texture and movement. But exactly what input each RGC sends to the brain, and how the brain uses this information, is unclear. Brackbill et al. set out to answer these questions by measuring and analyzing the electrical activity in isolated retinas from macaque monkeys. Studying the macaque retina was important because the primate visual system differs from that of other species in several ways. These include the numbers and types of RGCs present in the retina. These primates are also similar to humans in their high-resolution central vision and trichromatic color vision. Using electrode arrays to monitor hundreds of RGCs at the same time, Brackbill et al. recorded the responses of macaque retinas to real-life images of landscapes, objects, animals or people. Based on these recordings, plus existing knowledge about RGC responses, Brackbill et al. then attempted to reconstruct the original images using just the electrical activity recorded. The resulting reconstructions were similar across all retinas tested. Moreover, they showed a striking resemblance to the original images. These results made it possible to comprehend how the light-response properties of each cell represent visual information that can be used by the brain. Understanding how macaque retinas work in natural conditions is critical to decoding how our own retinas process and convey information. A better knowledge of how the brain uses this input to generate images could ultimately make it possible to design artificial retinas to restore vision in patients with certain forms of blindness.