1070奈米飛秒雷射在肺癌3D病理上的技術應用

Abstract

近年來，肺癌成為國人致死率最高的癌症，不管是先天性的還是後天環境造成，肺癌對大家的影響都不容忽視，也因此想利用手邊的顯微鏡系統對肺癌的結構做深入的研究，希望對肺癌的診斷能做出貢獻。 肺癌佔最大比例的是肺腺癌，所以我大部分作的檢體都是肺腺癌的，而我用的顯微鏡系統的激發光源是1070奈米的脈衝雷射，採用的是多光子激發的技術，在解析度的方面有一定的把關，且對於厚樣品的穿透性也很強，如此一來在厚檢體的3D呈現上會有優勢，所以我便把目光投向了3D病理。傳統病理都是對檢體做切片去做診斷，但3D能帶來的資訊量一定能超過2D，像是資料連續性(如z方向的連續性，切片會破壞一部份組織使得拼湊出來的三維影像部分資訊缺失)，以及結構彼此的連結都能更清楚的呈現。在檢體處理上我用了透明化處理讓檢體變得更透明使深層的信號能輕易的被捕捉到，接下來對數個檢體做取樣，從中觀察到各種肺腺癌的結構與型態。在3D呈現上可以隨意對其做旋轉看不同角度的結構，也可以看不同角度的切面，一般傳統切片診斷只能固定看一個切面，3D取樣可以呈現出無限多種切面，不同的切面也會帶來不同的結構從而影響診斷，再來，3D結構也使我們得以去大概計算細胞的數量，最後能分析出不同肺腺癌型態的細胞密度差異去對腫瘤嚴重性做分級，也能說是對診斷帶來合理的量化而不單純靠感覺，這些都是傳統2D診斷難以做到的。 在過程中我還發現了神祕的信號可以表達出細胞質與一些肺部結構的型態，而這神秘的信號竟是平時我們視為雜訊的自發螢光，此信號或多或少對實驗有幫助，因為它本來就存在，所以不用特別去染，在檢體處理上得以簡化步驟，再者，它均勻地存在於檢體之中，因此信號分布很漂亮。我也對該信號做了一些分析以確保它能勝任海量3D信息的表達，從實驗結果來看，它在3D呈現上資訊量還算完整。最後我還對3D影像做一些優化使解析度進一步的提升盡可能去滿足3D上各個角度的觀賞性，畢竟這也是3D呈現與生俱來的優勢之一，我要盡可能地放大它。

In recent years, lung cancer has become the leading cause of cancer mortality in Taiwan. Regardless of whether it stems from genetic factors or environmental influences, its impact cannot be ignored. Consequently, I intend to utilize my microscope system to conduct an in-depth study of lung cancer structures, aiming to contribute to its diagnosis. Lung adenocarcinoma accounts for the largest proportion of lung cancer cases, so most of the specimens I work with are of this type. My imaging system utilizes a 1070 nm pulsed laser as the excitation source and employs multiphoton excitation technology. This ensures high resolution and excellent penetration through thick samples, providing a significant advantage for 3D visualization. Therefore, I have focused my research on 3D pathology. Traditional pathology relies on tissue sectioning for diagnosis; however, the information provided by 3D imaging far exceeds that of 2D. 3D imaging preserves data continuity (e.g., continuity along the Z-axis, which is often lost in 2D sections due to tissue damage during slicing) and more clearly illustrates the interconnections between structures. To facilitate this, I applied tissue clearing techniques to enhance transparency, allowing signals from deeper layers to be easily captured. By sampling multiple specimens, I observed various structures and morphologies characteristic of lung adenocarcinoma. In a 3D representation, structures can be rotated to be viewed from any angle, and an infinite variety of cross-sections can be examined. While traditional biopsy diagnosis is limited to a single fixed plane, 3D sampling allows for diverse perspectives that reveal different structural details, influencing diagnostic outcomes. Furthermore, 3D structures enable us to estimate cell counts and analyze differences in cell density across different lung adenocarcinoma patterns. This provides a reasonable quantification for grading tumor severity rather than relying solely on subjective assessment—capabilities that are difficult to achieve with traditional 2D diagnosis. During this process, I discovered a "mysterious signal" that reveals the morphology of the cytoplasm and certain lung structures. This signal turned out to be autofluorescence, which is often dismissed as noise. This signal is beneficial to the experiment because it is endogenous, meaning no special staining is required, thereby simplifying sample preparation. Moreover, it is uniformly distributed throughout the specimen, resulting in high-quality signal distribution. I analyzed this signal to ensure its competency in representing massive 3D datasets; the results indicate that it provides comprehensive information for 3D visualization. Finally, I optimized the 3D images to further enhance resolution, maximizing the aesthetic and functional quality of the images from all angles—a natural advantage of 3D presentation that I aim to fully exploit.