透過您的圖書館登入
IP:54.227.104.229
  • 學位論文

自動化螢光顯微影像之次細胞結構辨識

Automated Subcellular Structure Recognition in Fluorescence Microscopy Images

指導教授 : 蔡育秀 林崇智
若您是本文的作者,可授權文章由華藝線上圖書館中協助推廣。

摘要


一般而言觀察次細胞結構及蛋白質型態須仰賴研究人員以目視的方法來比對判斷所取得之細胞影像,辨識正確率與研究人員的經驗成正比。本研究以數位影像處理的方式擷取螢光細胞影像的幾何特徵,再運用統計分析建立決策分類樹萃取辨識法則,其目的在建立一套自動化的螢光顯微影像辨識系統,來縮短研究人員的訓練與實驗後需觀察樣本的時間,以提升相關生物科技之研究效率。 研究方法是將影像先作比例與灰階度的正規化,再將影像二值化處理,最後抽取出二值化影像上的幾何特徵,利用統計分析來學習樣本的特徵數據,建立出決策分類樹來做為鑑別法則。 研究結果以國內CSMU(中山醫學大學)所提供八種次細胞結構(細胞核、細胞核仁、高爾基體、過氧化?體、粒腺體、肌動蛋白、微小管、內質網)學習建立決策分類樹後,對全部樣本八百七十五張影像作辨識,辨識率有90.6%的辨識正確率。使用國內所建立的決策樹來測試國外EMBL(德國海德堡的歐洲分子生物實驗室)所提供五種次細胞結構(細胞核、高爾基體、粒腺體、微小管、內質網)三百一十九張影像測試有78.4%的辨識正確率。而將國內外樣本混合再學習重建決策分類樹後,對於混合樣本一千一百九十四張的辨識率可達86.6%的辨識正確率,證明系統已有輔助實驗參考的價值。

並列摘要


In general, distinguishing subcellular structures and protein types in fluorescent microscopy images is based on researchers’ visual inspection. This task is an experience intensive job. The research is to apply geometry feature extraction and decision tree classification techniques in subcellular recognition. The goal is to setup an automated image recognition system to shorten the process time on post-experiment stage and accelerate research efficiency. The acquired microscopic images were first passed through scale and gray level normalization, bi-level processes. The geometric feature extraction routines were followed to analysis the spot size, number, relative distance, and characteristics, etc., from the bi-level images. These features were fed into statistic analysis procedures for decision tree construction. Total 1194 images collected from Chung Shan Medical University (CSMU) and the European Molecular Biology Laboratory (EMBL) were used to test the performance of the system. The results shown that the decision tree trained with the images offered by CSMU can achieve 90.6% correctness in distinguishing Nucleus, Nucleolus, Golgi, Peroxisome, Mitochondrial, Actin, Microtuble, and ER, on 875 images. When tested the decision tree with 319 EMBL’s images, it can offer 78.4% correctness in distinguishing Nucleus, Golgi, Microtuble, Mitochondrial and ER subcellular structures. When trained with the 240 mixture of CSMU and EMBL’s images, the decision tree can reach 86.6% correctness on all 1194 images.

參考文獻


[02] Skolnick J, Fetrow JS. “From genes to protein structure and function: novel applications of computational approaches in the genomicera. Trends”. Biotechnol 18(2000):34-9
[03] Boland M. V., Markey M. K. and Murphy R. F. “Automated Recognition of Patterns Characteristic of Subcellular Structures in Fluorescence Microscopy Images”. Cytometry 33(1998):366-375.
[04] Markey M. K., Boland M. V. and Murphy R. F. “Towards Objective Selection of Representative Microscope Images”. Biophys,76(1999): 2230-2237
[06] M. V. Boland and R. F. Murphy. “A Neural Network Classifier Capable of Recognizing the Patterns of all Major Subcellular Structures in Fluorescence Microscope Images of HeLa Cells”. Bioinformatics 17(2001) 1213-1223.
[07] Zhang J. and Tan T. “Brief review of Invariant Texture Analysis Methods”. Pattern Recognition 35(2001):735-747

被引用紀錄


陳佳駿(2016)。以蛋白質形態分佈建構內質網基因分群之研究〔碩士論文,中原大學〕。華藝線上圖書館。https://doi.org/10.6840%2fcycu201600206
但漢瑋(2011)。以蛋白質分佈之顯微形態結構進行嶄新內質網基因分類〔碩士論文,中原大學〕。華藝線上圖書館。https://doi.org/10.6840%2fcycu201100894
李玫憶(2005)。細胞次結構影像辨識系統〔碩士論文,中原大學〕。華藝線上圖書館。https://doi.org/10.6840%2fcycu200500778

延伸閱讀