請用此 Handle URI 來引用此文件:
http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/20164
標題: | 全域式光學同調斷層掃描儀於皮膚組織之量化參數研究 Quantitative analyses of human skin using full-field optical coherence tomography |
作者: | Chia-Kai Chang 張家凱 |
指導教授: | 黃升龍(Sheng-Lung Huang) |
關鍵字: | 光學同調斷層掃描儀,影像分析,隨機射線取樣,單細胞分割,組織特徵, optical coherence tomography,image analysis,random rayburst sampling,single-cell segmentation,tissue characterization, |
出版年 : | 2018 |
學位: | 博士 |
摘要: | 以往的生物細胞分析是透過二維影像來建立量化資訊,我們通過解析度為橫向0.8 µm,縱向0.9 µm的全域式同調斷層掃描(FF-OCT)系統,獲取三維斷層掃描立體影像,以三維立體資訊來分析,可以得到較全面的資訊。本研究開發隨機射線取樣(Random rayburst sampling; RRBS)框架,用於檢測三維空間中的細胞核與細胞膜的邊界,從而進行單細胞的量化分析,計算出單細胞的體積核質比(nuclear-to-cytoplasmic ratio)與細胞立體形貌。此RRBS框架的設計原則是透過隨機的取樣來得到最終的穩定客觀結果,因此可以減少人工操作的不便並避免人為干預。RRBS對雜訊不敏感,不同隨機選擇的RRBS計算所得到的體積核質比之間的相對標準偏差為2%。以螢光共焦顯微鏡做驗證,RRBS算法於體積核質比的誤差僅為1%。經模擬計算,若以二維影像來確認細胞的體積核質比,需要500顆細胞的橫切面才可得到穩定的統計數據(%RSD < 5%)。本研究透過三維斷層掃描技術與RRBS框架,可以測量並計算每顆個別細胞的體積核質比。為了提升計算效率,本研究使用圖形處理器(Graphics processing unit; GPU)的平行計算,若使用Nvidia的GTX 1080圖形處理器,於三維雙邊濾波器之計算可以得到280倍的速度提升。
RRBS框架也可應用於活體皮膚組織的量化分析。角質形成細胞(Keratinocytes)之細胞核的體積由顆粒層(Stratum granulosum)到基底層(Stratum basale)逐漸變小,結合FF-OCT的快速三維斷層掃描與RRBS框架可以提取位於表皮層(Epidermis)不同深度的角質形成細胞之細胞核的三維形態特徵,進而對活體皮膚進行量化分析,同時,使用RRBS框架可分析活體皮膚組織的角質形成細胞核的三維空間分佈,角質形成細胞核在基底層的平均縱向和橫向直徑分別為5.1±0.34和7.2±0.74 µm,RRBS框架與本實驗室的細胞等級高解析度FF-OCT搭配,有潛力作為皮膚之疾病與癌症的早期診斷的有效方法。 In the past, the quantitative analysis of biological cell is to establish information via two-dimensional (2D) images. Now, raw volumetric images were acquired through a full-field optical coherence tomography (FF-OCT) system with submicron resolution—i.e. 0.8 µm in lateral and 0.9 µm in axial direction. With the tomographic volumetric image data, more comprehensive information could be acquired. A random rayburst sampling (RRBS) framework was developed to detect the nucleus and cell membrane boundaries in three-dimensional (3D) space for determining the volumetric nuclear-to-cytoplasmic (N/C) ratio and morphology of a single cell. The design principle of this RRBS framework is to obtain the stable and objective results through random sampling. Therefore, the RRBS framework could reducing the inconvenience of manual operation and avoiding human intervention. The RRBS framework was insensitive to the selection of seeds and voxel noise. The relative standard deviation of the N/C ratio between different randomly selected seed sets was only 2%. Compared with the fluorescence confocal microscopy, the error of our algorithm is 1%. Compare with 2D images, need more than 500 exactly identical cells to obtain a precise (%RSD < 5%) volumetric N/C ratio. In this research, the volumetric N/C ratio of every single cell measured by FF-OCT could be calculated by RRBS framework. To improve computational efficiency, the graphics processing unit (GPU) is used for parallel processing. By using Nvidia GTX 1080, we have 280 times speed up for 3D bilateral filter. The RRBS framework was applied to in vivo human skin tissue with quantitative studies of the spatial distribution and morphometry of keratinocytes. The volume size of the keratinocyte nuclei became smaller from the stratum granulosum to the stratum basale. The rapid measurement of FF-OCT combined with a random rayburst sampling framework for the extraction of nuclear morphological features could serve as an effective approach to acquiring essential quantitative information for in vivo human skin tissue. Also, the spatial distribution of keratinocytes nuclei from in vivo human skin tissue were analyzed by this segmentation algorithm. The average axial and lateral diameters of keratinocyte nuclei in the stratum basale were 5.1±0.34 and 7.2±0.74 µm, respectively. The RRBS framework working with the high-resolution home-made OCT could serve as an effective approach to acquiring essential quantitative information for early stage disease diagnosis. |
URI: | http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/20164 |
DOI: | 10.6342/NTU201800508 |
全文授權: | 未授權 |
顯示於系所單位: | 光電工程學研究所 |
文件中的檔案:
檔案 | 大小 | 格式 | |
---|---|---|---|
ntu-107-1.pdf 目前未授權公開取用 | 6.34 MB | Adobe PDF |
系統中的文件,除了特別指名其著作權條款之外,均受到著作權保護,並且保留所有的權利。