本文選題：雙向凝膠電泳 + 核模糊聚類算法��； 參考：《山東師范大學(xué)》2017年碩士論文

【摘要】：蛋白質(zhì)是生命活動(dòng)的基礎(chǔ),對(duì)其功能及表達(dá)水平等方面的研究具有重要的價(jià)值,而快速發(fā)展的蛋白質(zhì)組學(xué)成為生物系統(tǒng)研究的一個(gè)新方向。將蛋白質(zhì)有效的分離出來(lái)是對(duì)其表達(dá)模式分析的前提,它是利用雙向凝膠電泳技術(shù)平臺(tái)來(lái)操作運(yùn)行,依靠蛋白質(zhì)在凝膠中水平維度上等電點(diǎn)的不同和豎直維度上分子量之間的差異特征實(shí)現(xiàn),最終輸出的是一幅數(shù)字灰度圖像。在這幅圖像中含有成千上萬(wàn)個(gè)形狀和大小各不相同的蛋白質(zhì)點(diǎn),其中每個(gè)點(diǎn)代表了一個(gè)蛋白質(zhì),并且這些點(diǎn)都是以不同灰度級(jí)的形式表現(xiàn)出來(lái)。凝膠圖像的分析過(guò)程主要包括:蛋白點(diǎn)的分割與檢測(cè),蛋白點(diǎn)的表達(dá)量化、匹配等過(guò)程,其中,蛋白點(diǎn)分割占據(jù)著非常重要的地位,不準(zhǔn)確的分割會(huì)嚴(yán)重影響后續(xù)對(duì)蛋白質(zhì)表達(dá)變化的分析,因此,本論文重點(diǎn)研究了對(duì)凝膠圖像的分割算法。本論文的主要工作如下:(1)研究了不同的濾波技術(shù)主要包括空域?yàn)V波、改進(jìn)的NL-means算法和引導(dǎo)濾波器,概述了其相關(guān)原理并分別對(duì)凝膠圖像進(jìn)行實(shí)驗(yàn)仿真。同時(shí),與處理后的凝膠圖像剖面圖相結(jié)合,進(jìn)一步對(duì)上述濾波算法處理的效果做出對(duì)比分析,并作為選擇濾波算法的依據(jù)。(2)將傳統(tǒng)的圖像分割算法(基于閾值、分水嶺和水平集算法)和基于模糊聚類的算法應(yīng)用于凝膠圖像的分割中,并利用真實(shí)的凝膠圖像對(duì)上述幾種算法進(jìn)行實(shí)驗(yàn),通過(guò)對(duì)比觀察實(shí)驗(yàn)結(jié)果來(lái)分析各種算法的分割效果,為下一步的研究提供基礎(chǔ)。(3)凝膠圖像上蛋白點(diǎn)并不是均勻分布且部分點(diǎn)的邊界灰度與背景之間的對(duì)比并不是特別明顯,由此將核模糊聚類算法引入到凝膠圖像的分割中并對(duì)其進(jìn)行改進(jìn)。首先將引導(dǎo)濾波器與形態(tài)學(xué)方法結(jié)合起來(lái),一方面用于對(duì)圖像進(jìn)行降噪處理以降低噪聲的干擾,另一方面用于提高圖像中蛋白點(diǎn)與背景之間的對(duì)比差異;然后在核函數(shù)中引入一個(gè)權(quán)值向量,與此同時(shí)利用樣本方差來(lái)合理的計(jì)算核參數(shù)值,使核參數(shù)具有一定的自適應(yīng)度;最后將改進(jìn)后的核函數(shù)引入到模糊聚類算法里面,從而最終完成凝膠圖像的聚類分割。在此過(guò)程中,分別利用模擬和真實(shí)的凝膠圖像進(jìn)行實(shí)驗(yàn),并與其他算法做對(duì)比分析,驗(yàn)證了本論文算法的分割效果明顯優(yōu)于對(duì)比算法,在一定程度上能夠分割出更多的微弱蛋白點(diǎn),提高了凝膠圖像分割的精度和準(zhǔn)確性。
[Abstract]:Protein is the basis of life activity, so it is of great value to study its function and expression level. The rapid development of proteomics has become a new direction of biological system research. The efficient separation of proteins is a prerequisite for the analysis of their expression patterns, which are operated on a two-dimensional gel electrophoresis platform. Depending on the difference of isoelectric point in the horizontal dimension and the molecular weight in the vertical dimension of the gel, the final output is a digital gray image. The image contains thousands of protein spots of different shapes and sizes, each of which represents a protein, and these dots are represented by different grayscale levels. The analysis process of gel image mainly includes: protein point segmentation and detection, protein point expression quantification, matching and so on. Among them, protein point segmentation occupies a very important position. Inaccurate segmentation will seriously affect the subsequent analysis of protein expression changes. Therefore, this paper focuses on the segmentation algorithm of gel image. The main work of this thesis is as follows: (1) the main filtering techniques include spatial filtering, improved NL-means algorithm and bootstrapping filter. At the same time, combining with the processed gel image profile, the effect of the above filtering algorithm is compared and analyzed, and the traditional image segmentation algorithm (based on the threshold value) is used as the basis for selecting the filtering algorithm. Watershed and level set algorithm) and fuzzy clustering algorithm are applied to the segmentation of gel image, and the real gel image is used to experiment the above algorithms. The segmentation effect of these algorithms is analyzed by comparing and observing the experimental results. To provide the basis for further research.) the protein spots on the gel image are not uniformly distributed and the contrast between the boundary grayscale of some points and the background is not particularly obvious. Therefore, kernel fuzzy clustering algorithm is introduced into gel image segmentation and improved. Firstly, the guided filter is combined with the morphological method, which is used to reduce the noise interference of the image, on the one hand, to improve the contrast between the protein points and the background in the image. Then a weight vector is introduced into the kernel function, and at the same time, the kernel parameter value is reasonably calculated by using the sample variance, so that the kernel parameter has a certain degree of adaptability. Finally, the improved kernel function is introduced into the fuzzy clustering algorithm. Finally, the clustering segmentation of gel image is completed. In this process, simulation and real gel images are used to carry out experiments, and compared with other algorithms, it is verified that the segmentation effect of this algorithm is obviously better than that of contrast algorithm. To some extent, more weak protein points can be segmented, and the accuracy and accuracy of gel image segmentation are improved.
【學(xué)位授予單位】：山東師范大學(xué)
【學(xué)位級(jí)別】：碩士
【學(xué)位授予年份】：2017
【分類號(hào)】：Q51;TP391.41

【參考文獻(xiàn)】

相關(guān)期刊論文 前4條

1 張彥清;侯偉;李玲;辛化梅;;基于改進(jìn)MRF的凝膠圖像分割算法[J];科技創(chuàng)新與應(yīng)用;2014年07期

2 李赫;鄒黎明;;雙向電泳圖譜的圖像分析技術(shù)進(jìn)展[J];沈陽(yáng)醫(yī)學(xué)院學(xué)報(bào);2006年01期

3 畢于慧,崔賽華;基于分水嶺算法的雙向凝膠電泳圖像分割[J];內(nèi)蒙古林業(yè)科技;2004年04期

4 張莉,周偉達(dá),焦李成;核聚類算法[J];計(jì)算機(jī)學(xué)報(bào);2002年06期

相關(guān)博士學(xué)位論文 前4條

1 方江雄;基于變分水平集的圖像分割方法研究[D];上海交通大學(xué);2012年

2 李曉敏;大腸桿菌氧脅迫適應(yīng)性蛋白質(zhì)組學(xué)、O-抗原研究和嗜熱菌NG80-2醛脫氫酶分子生物學(xué)研究[D];南開(kāi)大學(xué);2010年

3 黃冰心;基于凝膠電泳的中國(guó)明對(duì)蝦血漿蛋白質(zhì)組學(xué)研究[D];中國(guó)科學(xué)院研究生院（海洋研究所）;2008年

4 黃珍玉;動(dòng)脈粥樣硬化疾病模型蛋白質(zhì)組及其方法學(xué)研究[D];復(fù)旦大學(xué);2005年

相關(guān)碩士學(xué)位論文 前5條

1 趙風(fēng)平;蛋白質(zhì)凝膠圖像消噪技術(shù)研究[D];山東師范大學(xué);2012年

2 王峰;圖像引導(dǎo)濾波器在單幅圖像復(fù)原中的應(yīng)用[D];合肥工業(yè)大學(xué);2012年

3 王延翔;基于均值算法的混合噪聲圖像濾波算法的研究與實(shí)現(xiàn)[D];北京郵電大學(xué);2010年

4 劉祝華;圖像去噪方法的研究[D];江西師范大學(xué);2005年

5 楊國(guó)勝;凝膠電泳圖像計(jì)算機(jī)分析與實(shí)驗(yàn)研究[D];河北工業(yè)大學(xué);2002年

，

本文編號(hào)：1796237