Unlocking the potential of High-speed widefield Imaging

项目总结： 宽视场成像和空间多路复用对于推进 神经科学。目前的成像器不能提供所需的速度和多功能性 钙离子或电压成像实验。在终身成像的情况下 Cmos成像器完全缺乏功能。这个问题更加微妙 因为这不仅仅是暴力加速的问题 技术速度的提高伴随着大量的功率消耗和 对更快的数据接口的需求。一种成像技术有可能 解决这一问题的是单光子雪崩探测器(SPAD)。以德为本的铁锹 操作的结果是数字脉冲实际上消除了 通常困扰着普通的成像者。然而，到目前为止，SPAD成像器还没有 由于像素密度较低，因此被广泛使用。最大的优势之一是 SPAD成像器具有执行时间相关测量的能力，使 终身成像。寿命成像仪可以产生绝对的定量测量结果 常规的成像方式是不可能的。然而，目前的SPAD 成像器需要几秒钟到几分钟的时间来计算一张终生图像，对于 神经成像。我们建议通过以下方式克服这些根本障碍 在设备、架构和封装层面进行创新。我们的建议 该方法利用晶体管放大的SPAD设计与像素模拟相结合 计数和寿命估算。这两项创新使阅读成为可能 像素级数据以较慢的速率传输，同时保持较快的帧速率。我们 此外，还提出了一种新的芯片级集成方法，该方法将 硅封装中的成像器管芯和处理管芯 子阵列。这种方法使成像器能够保持 随着像素密度的缩放，像素子阵列。最后，我们将演示 我们的成像仪通过成像树突活动，在强度和 在史无前例的时空神经培养中的终身成像模式 比例。

Project Summary: Widefield imaging and spatial multiplexing are crucial to advancing the field of neuroscience. Current imagers do not offer the speed and versatility needed for calcium or voltage imaging experiments. In the case of lifetime imaging the functionality is completely lacking in CMOS imagers. The problem is more subtle than it seems because it is not just a matter of brute force speed-up through technology. Speed increases come with large amounts of power dissipation and the need for faster data interfaces. One imaging technique with the potential to solve this issue is the single photon avalanche detector (SPAD). SPADs by virtue of operation result in digital pulse practically eliminating read noise that commonly plagues regular imagers. However, SPAD imagers till date have not seen widespread use due to low pixel density. One of the greatest advantages of the SPAD imager is its ability to perform time correlated measurements, enabling lifetime imaging. Lifetime imagers can yield absolute quantitative measurements not possible with regular imaging modalities. However, the current SPAD imagers take seconds to minutes to compute a lifetime image, much too slow for neural imaging. We propose to overcome these fundamental barriers by innovating at the device, architecture and packaging levels. Our proposed approach utilizes a transistor amplified SPAD design coupled with in pixel analog counting and lifetime estimation. These two innovations make it possible to read the pixel level data at a slower rate, while maintaining a fast frame rate. We additionally propose a new chip level integration approach which packages the imager die and processing die in a silicon package enabling reading from subarrays. This approach enables the imager to maintain the frame rate of the pixel subarray as the pixel density scales. Finally, we demonstrate the advantages of our imager by imaging dendritic activity, both in intensity and lifetime imaging modes, in neural cultures at unprecedented spatiotemporal scales.