高速三維立體影像顯微鏡技術

High Speed 3D Volume Imaging Microscopy

林瑋冠(Wei-Kuan Lin)

指導教授 ： 朱士維

國立臺灣大學/理學院/物理學研究所/碩士(2016年)

https://doi.org/10.6342/NTU201603141

摘要

了解大腦是如何運作的對於腦神經科學家而言是最重要的研究目標之一。近幾年來，神經科學家們相信大腦內部的神經連結是使大腦形成功能的最基礎單元，而這些神經連結被稱為connectome。了解這些連結是如何傳遞訊號、和形成功能性神經網路可以幫助我們了解許多重要的問題，包含記憶是如何產生的，為什麼人類會有意識等等。
然而，要了解功能性神經網路並非一件簡單的事。在神經網路中，每一個神經元都有著複雜的三維結構，而神經元的大小都只有僅僅數微米大，且神經事件的產生都只有僅僅數毫秒的時間。此外，神經訊號的量測必須為非侵入式以避免環境改變而造成神經訊號的傳遞改變。因此，發展一個同時能夠記錄三維神經訊號、非侵入式且同時具有高空間解析度及時間解析度的工具有著高度的重要性。
在現有的技術中，光學顯微鏡是一個最可能符合上述條件的技術。光學顯微鏡是一個非侵入性的成像方法，能夠提供低於微米等級的解析度能力。此外，光學顯微鏡能夠量測許多神經的功能性反應，例如：鈣離子濃度變化、膜電位的改變等等。然而，光學顯微鏡在三維高解析度的成像速度上仍然有著侷限性，在深組織的成像上，光學顯微鏡必須仰賴雙光子掃描顯微鏡技術達成光學切片效果及減少散射的影響。但也因為雙光子掃描顯微鏡必須先在XY平面上掃描每個點後再慢慢移動物鏡或載物台逐層掃描而降低了光學顯微鏡的成像速率。
這幾年來有不少人提出改進雙光子顯微鏡成像的方法，然而大部分的方法都還是只能提升其中一個軸度的成像速度，使得速度上仍不足以快到跟上神經的傳遞速率。有些團隊成功同時提高了軸向和橫向的成像速度，例如：快速變焦鏡與光片照明顯微鏡的結合，但這類的方法往往依賴廣視野顯微鏡技術的使用，會造成很強的散射，使得在深組織影像中無法使用。
為了加快雙光子掃描顯微鏡的速度，我們自行設計一套多光子多焦點顯微鏡，提升橫向上的成像速率。在軸向上，我們利用了快速變焦透鏡達到快速焦距變化，進而達到高速掃描。這個顯微鏡的組合不僅可以同時提升軸向和橫向的成像速度，另外我們保留了掃描顯微鏡的特性，使得在深組織中至於因為訊號光的散射而不至於導致影像品質的變壞。
除此之外，在這篇論文裡我們也提出一個搭配高速掃描系統的超高速脈衝偵測系統之新設計超高速的光脈衝偵測系統用以提升系統訊雜比。其概念主要為利用光電倍增管對光子的反應具有特定波型和雷射脈衝具有固定週期之特性的以作為過濾雜訊的根據。我們設計了兩種方法：分別為對光電倍增管輸出對特定波型做擬合得出光強度大小之方法及利用time-gated方法，僅在螢光放光其間取值進行訊雜比之提升。
在本研究中，我們成功架設出了快速掃描顯微鏡系統及高速脈衝偵測系統。雖然目前還有一些問題如多焦點造成的串擾問題及偵測系統的雜訊問題，但我們仍然期待這套系統之後可以做為腦神經科學研究的重要工具。

關鍵字

多光子顯微鏡 ； 多焦點顯微鏡 ； 快速變焦透鏡

並列摘要

Understanding how the brain functions is a fundamental goal for neuroscientists. Over years of study, neuroscietists belive the neural interconnections insided the brain is the most fundamental thing to function the brain and the connections are are termed “connectomes.” The communication and functional connectivity of these connectomes are called “functional connectomes.” Resolving these functional connectomes can help people understand some essential questions, such as how the memories forms, and why we have consciousness
Nevertheless, functional connectome study is not an easy task. It is because that the neuron networks inside the brain exhibit three-dimensional structures. The spatial size of neuron is small on the order of several micrometer (μm) and the temporal dynamics of neural events is fast with several millisecond (ms) in scale. Besides, the functional neurons can hide deep from the brain surface, requiring the recording method to be not only non-invasive, but also capable of isolating from the environment. Therefore, to study functional connectome, it is necessary to develop tools with capability of three-dimensional recording, non-invasive observation and high spatail and temporal resolution.
Among current techniques, optical microscopy is a promising method to meet those requirements. It can achieve non-invasive detection and provide sub-μm spatial resolution. Besides, optical microscopy is compatible with a variety of functional sensors such as voltage, calcium, or metabolic indicators. Nevertheless, optical microscopy still has its limitation. In deep tissue imaging, conventional wide-field optical microscopy suffers from strong scattering and signal mix-up due to lack of optical sectioning capability. Although two-photon laser scanning microscopy provides significant improvement on these issues, however, the imaging speed is limited. To acquire a 3D volume image, it requires to laterally scan every pixel on XY plane and then scan axially by moving objective lens or sample.
During the past decade, several techniques have been proposed to enhance the imaging speed of two-photon microscopy. However, most of them cannot enhance the imaging speed on lateral and axial direction at the same time, making it still difficult to catch up the neuron signal propogation. Recently some groups have successfully enhance speed on both lateral and axial directions, such as the combination of light-sheet microscopy and a electric tunable lens. However, those techniques are wide-field illumination based and the signal may suffer from scattering.
In this thesis, to improve the slow 3D imaging prbolem, we combine multifocal multiphoton microscopy with a tunable acoustic gradient index of refraction lens to enhance lateral and axial scanning speed simultaneously. This combination can greatly enhance imaging speed as well as reduce the effect of tissue scattering due to its scanning feature.
Furthermore, we design and realize a high speed pulse detection system to improve signal-to-noise ratio, based on the knowledge that the pulse repsonse of photomultiplier tube has a charactertic waveform and the laser pulse repetiton rate is fixed. In our work, we propose two methods: one is signal pulse fitting method that can filter out the noises that doesn’t fit the signal waveform requirements. The other one is time-gated method, which detects signals only within a specific time after each excitation pulse.
By combining the high-speed scanning system with the high-speed detection system, we will be able to detect 3D neuronal dynamics inside brain with very high spatiotemporal resolution. Though currently, our techniques still require to be improved in terms of multiple beams cross-talk and noise influence, we expect this system will be a powerful tool for functional connectome study in the near future.

並列關鍵字

multifocal microscopy ； multiphoton microscopy ； fast tunable lens

參考文獻

1. Pelvig, D.P., et al., Neocortical glial cell numbers in human brains. Neurobiology of Aging, 2008. 29(11): p. 1754-1762.

2. Sporns, O., G. Tononi, and R. Kotter, The Human Connectome: A Structural Description of the Human Brain. PLoS Comput Biol, 2005. 1(4): p. e42.

3. Schrodel, T., et al., Brain-wide 3D imaging of neuronal activity in Caenorhabditis elegans with sculpted light. Nat Meth, 2013. 10(10): p. 1013-1020.

4. Briggman, K.L. and W.B. Kristan, Multifunctional Pattern-Generating Circuits. Annual Review of Neuroscience, 2008. 31(1): p. 271-294.

5. Niessing, J. and R.W. Friedrich, Olfactory pattern classification by discrete neuronal network states. Nature, 2010. 465(7294): p. 47-52.