基於全域式光學同調斷層掃描術進行三維動態成像開發

Abstract

光學同調斷層掃描術(Opticalcoherence tomography; OCT) 因其微米級的解析度、高速的掃描速度和非侵入性等特性而廣為人知。本研究整合了260幀率的高速CCD相機以及95奈米頻寬的寬頻光源於Ce3+:YAG全域式光學同調斷層掃描系統(full-field OCT; FF-OCT)，從而在進行非侵入式掃描時達到0.8微米的橫向解析度及1微米的軸向解析度。利用先前開發的動態FF-OCT(DynamicFF-OCT; D-FF-OCT) 掃描模式，我們可以擷取到待測樣本的動態資訊，並具備3.8微秒的高時間解析度。再透過D-FF-OCT程式的分析產生兩種二維的動態影像結果:一為與標準差值相關的動態光學同調斷層掃描影像(D-FF-OCTimage)，二為與頻譜資訊相關的上色動態光學同調斷層掃描影像(Color-codedD-FF-OCTimage)。

本研究進一步開發了三維D-FF-OCT成像的方法。在數據擷取方面，不同於二維D-FF-OCT 成像所使用的”固定深度掃描模式”(即動態FF-OCT掃描模式)，我們採用“極低速掃描模式＂讓PZT(Piezoelectrictransducer) 以每秒 0.05 微米的速度移動，使得在1微米樣本厚度內(相當於二維動態掃描模式所具有的面厚度)可擷取高達5200張的原始數據，達到如同”動態FF-OCT掃描模式”般的掃描效力。此外，對於使用慢速移動PZT所擷取到的三維動態成像原始數據，其不僅具有時間相關性，也具有些微的深度相關性。經本研究分析並設置適當的擷取參數後，三維動態分析結果不僅同樣可以呈現二維動態結果所得到的樣本動態資訊，更因其時間相關原始數據實際具有的面厚度，使得對時間資訊平均的同時也平均了空間資訊，從而提升了動態影像結果的信噪比。在程式分析方面，我們開發的三維動態分析程式不僅解決掃描三維動態原始資料的長時間問題，更具備處理大量三維原始資料的能力。此外，我們也進一步改善程式使其分析時間減半，並增強了最終上色動態影像的色彩對比度。

本研究使用人類角膜上皮細胞(Humancornealepithelial cells; HCECs) 作為三維D-FF-OCT 成像方法的測試樣本，並展示了其對細胞內細小結構（如染色質和TNT）的動態成像能力。此外，我們也呈現了從兔子口腔黏膜上皮細胞片獲得的橫平面(enface) 及縱平面(crosssection) 動態影像結果，說明三維D-FF-OCT成像方法能夠提供對於組織靜態(形態方面)和動態特性的全面了解。

Optical coherence tomography (OCT) has garnered acclaim for its micrometer resolution, high scanning speed, and non-invasive examination. This study harnesses a Ce³⁺:YAG full-field OCT (FF-OCT) system, integrating a high-frame-rate CCD sensor of 260 fps and broadband light of 95-nm spectrum bandwidth, to achieve non-invasive imaging with lateral and axial resolutions of 0.8 μm and 1 μm, respectively. Utilizing the 2D D-FF-OCT scanning mode, we capture time-dependent raw data with a temporal resolution of 3.8 ms, producing D-FF-OCT and color-coded D-FF-OCT images corresponding to standard deviation (STD) values and spectrum performance.

Furthermore, this study develops the methodology of the 3D D-FF-OCT technique. Employing an ”extremely low-speed scanning mode” with a PZT moving at 0.5 μm per second, we acquire up to 5200 slices within a 1-μm depth (equal to the plane thickness of 2D D-FF-OCT image), which is as stationary as the D-FF-OCT scanning mode. Moreover, the raw data of 3D D-FF-OCT captured with the extremely low-speed scanning mode is time-dependent and slightly depth-dependent. Through the analysis in this study, the 3D D-FF-OCT results can provide dynamic information within the sample as the 2D D-FF-OCT method does. The 3D D-FF-OCT method can also reach a better signal-to-noise ratio (SNR) due to the average processing conducted on time-dependent raw data (which is also depth-dependent). Our analysis program for 3D dynamic imaging addresses the challenges of long scanning times and extensive RAM usage associated with 3D D-FF-OCT raw data. We also halved the analysis time and enhanced the contrast of the final color-coded D-FF-OCT results.

As a demonstration, human corneal epithelial cells (HCECs) serve as the test sample for the 3D dynamic FF-OCT method, showcasing its dynamic-imaging capability of tiny or thin structures within cells, such as chromatin and TNT. Additionally, we present enface and cross-sectional dynamic results obtained from rabbit oral mucosa epithelial cell sheet, illustrating 3D D-FF-OCT’s capacity to offer comprehensive insights into tissue morphology and dynamic properties.