仿效兔子視網膜開持續中樞神經細胞之視覺神經晶片

Abstract

哺乳類動物視網膜由五種不同的細胞構成，而每一種細胞又各自有不同的分類與功能。除了將影像世界的訊息轉變為神經訊號外，視網膜還負責0.3%的大腦視覺處理功能。在本研究中，我們嘗試以工程的角度去瞭解視網膜的運作功能。 於本論文中，我們提出了一個新的設計方法用以設計金氧半場效電晶體神經型態晶片，此神經型態晶片依據由哺乳類動物視網膜建立之生物模型所設計的仿視網膜矽晶片，模擬了兔子視網膜的開持續中樞神經節細胞組；文中詳論其特性與工作原理。 此晶片包含一個大小為32x32的影像感測陣列，可仿效兔子視網膜開持續中樞神經細胞的功能。陣列中包含了許多相同的基本單元，而每個基本單元中包含了光輸入元件、感光細胞、水平細胞、開兩極細胞、關兩極細胞、Amacrine細胞與中樞神經細胞元件。晶片中還包括了輸入輸出接墊（I/O pads）的靜電防護、位址解碼（address decoder）及訊號讀出電路。 最後，我們比較原本生物模型的結果與電路模擬及晶片的量測結果之異同，檢驗此晶片的正確性與待改進之處。量測結果顯示，該金氧半場效電晶體神經型態晶片在時域及空間域的特性與生物量測結果相符，驗證了該晶片的仿生功能。經由實驗驗證，這個設計方法所設計的神經型態晶片能協助尚未揭露的視網膜細胞行為與視覺語言，並且也可用以設計所有的視網膜神經節細胞組，將視網膜功能完全由晶片實現。此外，該研究結果也使得極具潛力的視網膜晶片應用，諸如運動偵測、電腦視覺、人工視網膜和生醫元件方面變的更有可行性。

The mammalian retina is comprised of five different kinds of cells, for each kind can be divided into more types. Besides transducing the visual world into neural signals, these cells are in charge of 0.3% of the visual function of the brain. In this research, we try to understand the retinal functions from the engineering point of view. In this thesis, a new design methodology is proposed to implement CMOS neuromorphic chips which imitate the ON sluggish sustain ganglion cell set of rabbits’ retinas. A retinal chip based on the biological model derived from the mammalian retina is proposed. The retinal chip contains a focal plane sensory array of 32x32 similar basic cells that perform the functions. Each basic cell contains photo-input, photoreceptor, horizontal cell, on bipolar cell, off bipolar cell, amacrine cell and ganglion cell. ESD protection circuit for I/O pads, address decoders and readout circuits are also included in this chip. Finally, the results of HSPICE simulation and chip measurement are compared with that of the original model to examine the consistency and the difference for further improvement. The measurement results on the fabricated CMOS neuromorphic chip are consistent with the biological measurement results. Thus the biological functions of the chip have been successfully verified. It can be used to understand more biological behaviors and visual language of retina under different input optical images which have not yet been tested in biological experiments. Based on the results, the full ganglion cell sets of retina can be designed. Otherwise, many potential applications of retinal chips on motion sensors, computer vision, retinal prosthesis, and biomedical devices are feasible.