高速三維立體影像顯微鏡

Abstract

大腦是控制人類生理、行為、情緒、意識、記憶及其他一切生命徵象等人類活動的重要器官。儘管人類對於腦神經科學的研究，從對個別單神經細胞的行為反應開始至今已逾百年，但人類對於大腦複雜功能的認識仍然非常有限。其中部分關鍵在於，大腦複雜的功能並非由單一腦神經細胞或一叢神經聚落形成，而是由神經細胞彼此間複雜的連結而成，因此近年有許多出色的研究探討神經細胞間功能複雜的連結網路。然而近年多數的研究透過解剖學的角度探討神經細胞間的結構與區塊連結，雖然功能性神經連結能對大腦功能的運作提供更多資訊，卻受限於技術的限制而難以有足夠的時間、空間解析度應用於全腦功能的研究。為了能夠研究活體內大腦功能性神經連結的運作機制，需要發展一套技術具有高空間解析度(~μm)以分辨單神經細胞的活動、高速成像能力(~ms)以追蹤每一個神經動作電位訊號、接近毫米尺度的立方體成像範圍以完整記錄腦神神經連結的資訊。 綜觀近來的技術，光學顯微鏡是最佳能夠符合上述條件的技術。光學顯微鏡是非侵入性的成像方法且能提供次微米尺度的空間解析度。此外，光學顯微鏡能與許多種不同的標記方法相容，例如膜電位變化、鈣離子濃度變化等標記方法。然而，光學顯微鏡在三維高解析度的成像速度上仍然有其限制，在深組織的成像上也受到組織散射影響成像品質，並且缺乏縱向光學切片效果的能力。雖然雙光子雷射掃描顯微鏡能夠改善上述問題，卻因其需要掃描二維的每個像素點而後移動物鏡或載物平台而降低了成像速率。 數十年來，有許多提升三維成像速度應用於神經功能連結的技術發展，包括壓電位移平台、聲光透鏡、空間光調變器、光場顯微鏡、光片顯微鏡、液晶透鏡等，然而大多數的方法不易同時提升橫向與縱向的成像速度，亦或是因其廣視野的成像方法而仍會受到組織散射的影響而僅能用於透明樣品中。為了解決這個問題，我們已經發展一套有效的技術，結合高速縱向掃描透鏡及多光子掃描顯微鏡，達成在半秒的尺度內完成高速三維立體影像。 然而在先前的系統中，雖然縱向掃描速度可以超過100kHz，但立體成像的速度仍然被焦點於二維掃描時的時間所限制。故在此篇論文中，我們發展了一套32道光束的多焦點掃描顯微鏡系統，大幅提升了原先的橫向掃描速度至一到二個數量級。藉由多焦點掃描系統與快速縱向掃描透鏡的結合，我們可以同時提升橫向及縱向的掃描速度，並達成毫秒尺度的高速三維立體影像，符合我們對於量測神經細胞動作電位的要求。除此之外，為了正確即時取得高速影像數據的結果，我們與南方科技公司合作開發了一套高速資料擷取系統，以快速地完成資料串流儲存與正確的三維影像重組。在本篇論文中，我們成功地證明這套系統能符合我們預期的目標，並且在不久的將來就能作為研究活體腦神經功能連結網絡的重要技術。

Brain, which controls our behaviors, emotions, consciousness, memories and all other vital signs of human, is one of the most important organs in human body. Although the history of studying individual neuron behaviors is more than 100 years, our understanding toward brain function is still very limited. One of the key reasons is that brain function is not dictated by a single neuron, but emerges from the sophisticated connections among neurons, i.e. connectome. Therefore, recently there have been significant amount of researches toward the understanding of connectome. However, most current studies concentrate on “structural” connectome through anatomical approaches, while the development of “functional connectome”, though much more informative, is limited by technical difficulties to map the brain with high enough spatiotemporal resolution throughout the whole volume. In order to study the way how functional connectome works in brain of living animals, the required tool should exhibit spatial resolution less than ~μm, temporal imaging speed as high as ~ms, and imaging size approaching millimeter cubic in volume. Among current techniques, optical microscopy is the best candidate to match these specifications. It can achieve non-invasive detection and provide sub-μm spatial resolution. In addition, optical microscopy is compatible with a variety of probes such as voltage, calcium or metabolic indicators. However, optical microscopy still has its limitation. In deep tissue imaging, conventional wide-field optical microscopy suffers from strong scattering and lack of optical sectioning capability. Although two-photon laser scanning microscopy provides siginificant improvement on the issues, the imaging speed is limited. To acquire 3D volumetric image, it have to scan every pixel laterally on xy plane and then scan axially by moving objective lens or sample. During the past decade, several schemes have been demonstrated recently to boost 3D imaging speed for functional connectome study of brain in vivo, including piezoelectric translator, acousto-optic deflectors, spatial light modulator, light field microscopy, light sheet microscopy, liquid lens, etc. However, most of these techniques cannot enhance imaging speed on both lateral and axial directions simultaneously. In addition, some of these suffer from scattering from tissue because of its wide-field illumination or can only use for transparent samples. To overcome these issues, we have presented a powerful technique that combines high-speed axial scanning lens and multiphoton microscopy to increase scanning speed of 3D volumetric imaging down to sub-second scale. However, in the earlier system, even though the axial scanning speed is more than 100 kHz, the 3D volume imaging speed is limited by the single point scanning on the lateral plane. In this thesis, we developed a 32-beam multifocal system to significantly enhance the lateral scanning speed with 1-2 orders of magnitude. Combined with the high-speed axial scanning lens, the volumetric imaging speed approaches milliseconds scale, that is adequate to probe dynamics of action potential among neurons. In addtion, in order to acquire large amount of signal from fast imaging, we cooperate with SouthPort Co.Ltd. Taiwan to achieve high-speed data streaming and 3D image reconstruction. In this thesis, we present preliminary results to demonstarte the system performance. This powerful and innovative system is expected to greatly facilitate in vivo brain functional connectome studies in the near future.