本文選題：卷積神經(jīng)網(wǎng)絡(luò) + 醫(yī)學(xué)圖像處理��； 參考：《湖北工業(yè)大學(xué)》2017年碩士論文

【摘要】：卷積神經(jīng)網(wǎng)絡(luò)對(duì)圖像分類問(wèn)題的處理往往優(yōu)于其他同類型算法,其中的卷積層和子采樣層具有能夠提取樣本特征的功能,而共享權(quán)值的特點(diǎn)又極大減少了網(wǎng)絡(luò)需要訓(xùn)練的參數(shù)�？萍疾粩噙M(jìn)步的今天,醫(yī)療技術(shù)也得到了飛快的發(fā)展,從中產(chǎn)生的各種病癥檢查圖片更是數(shù)不勝數(shù)。醫(yī)師急需擺脫各種繁重的醫(yī)學(xué)圖像篩查工作,且如何從無(wú)數(shù)的病例圖中找出某種疾病的相似特征等;如此種種困難不斷激勵(lì)著研究人員,醫(yī)學(xué)圖像的研究也漸漸的成為了熱點(diǎn)。本文對(duì)卷積神經(jīng)網(wǎng)絡(luò)應(yīng)用于兩類醫(yī)學(xué)圖像展開了研究,其中一類為能反映身體疾病的眼球血絲圖,另一類為含有各種級(jí)數(shù)的腦膠質(zhì)瘤核磁共振成像圖。全文工作如下:(1)首先介紹了卷積神經(jīng)網(wǎng)絡(luò)的發(fā)展歷程,包括國(guó)外與國(guó)內(nèi)對(duì)其研究的成果,并且詳細(xì)的說(shuō)明了卷積神經(jīng)網(wǎng)絡(luò)的結(jié)構(gòu),算法以及推導(dǎo),完整的闡述了復(fù)雜的圖像分類問(wèn)題中應(yīng)用卷積神經(jīng)網(wǎng)絡(luò)的優(yōu)越性。(2)在經(jīng)典LeNet-5卷積神經(jīng)網(wǎng)絡(luò)結(jié)構(gòu)上實(shí)施改進(jìn),設(shè)計(jì)了具有不同卷積核,不同子采樣方式與不同分類器的網(wǎng)絡(luò)結(jié)構(gòu),并把此結(jié)構(gòu)用于解決識(shí)別眼球血絲病癥問(wèn)題。同時(shí)在實(shí)驗(yàn)環(huán)節(jié)對(duì)輸入層樣本尺寸,網(wǎng)絡(luò)的迭代次數(shù)進(jìn)行了探究,對(duì)比了改進(jìn)結(jié)構(gòu)與LeNet-5在使用同一樣本數(shù)據(jù)集情況下的區(qū)別,實(shí)驗(yàn)表明改進(jìn)結(jié)構(gòu)能很好的分類眼球血絲所反映的病癥。(3)根據(jù)腦膠質(zhì)瘤多層圖片的特點(diǎn),并基于眼球血絲網(wǎng)絡(luò)模型,設(shè)計(jì)出多列卷積神經(jīng)網(wǎng)絡(luò)結(jié)構(gòu):每一層的腦膠質(zhì)瘤樣本作為每一列的輸入,同時(shí)增加了卷積和子采樣層的層數(shù),并使用Maxout激活函數(shù)替代了傳統(tǒng)神經(jīng)網(wǎng)絡(luò)中經(jīng)常使用的Sigmoid函數(shù)。實(shí)驗(yàn)部分取多列結(jié)構(gòu)與單列結(jié)構(gòu),人工提取特征方式實(shí)施了對(duì)比,結(jié)果凸顯了多列結(jié)構(gòu)在膠質(zhì)瘤分級(jí)上的優(yōu)勢(shì);此外還對(duì)樣本進(jìn)行優(yōu)化處理,進(jìn)一步提高了分級(jí)準(zhǔn)確度。最后本文對(duì)多列卷積神經(jīng)網(wǎng)絡(luò)計(jì)算進(jìn)行了可視化處理,從視覺(jué)方面解釋了每層的工作過(guò)程。
[Abstract]:Convolution neural network is usually superior to other similar algorithms in image classification problem. The convolution layer and sub-sampling layer can extract the feature of samples, and the characteristics of shared weights greatly reduce the network parameters that need to be trained. With the development of science and technology, medical technology is developing rapidly. Doctors urgently need to get rid of all kinds of heavy medical image screening work, and how to find out the similar characteristics of a disease from countless case maps, and so on; such difficulties continue to inspire researchers, medical image research has gradually become a hot spot. In this paper, the application of convolutional neural networks to two kinds of medical images is studied. One is a hemodigram of the eyeball which can reflect the body disease, the other is a magnetic resonance imaging of glioma with various stages. The main work of this paper is as follows: (1) the development of convolutional neural network is introduced, including the research results both at home and abroad, and the structure, algorithm and derivation of convolutional neural network are explained in detail. The advantages of applying convolution neural network in complex image classification problem are discussed. (2) the network structure with different convolution kernel, different subsampling method and different classifier is designed by improving the classical LeNet-5 convolutional neural network structure. And this structure is used to solve the problem of identifying ocular blood disease. At the same time, the sample size of the input layer and the number of iterations of the network are explored in the experiment, and the difference between the improved structure and LeNet-5 in the case of using the same sample data set is compared. The experimental results show that the improved structure can well classify the diseases reflected by the blood filaments of the eyeball. (3) according to the characteristics of the multilayer images of gliomas and based on the network model of the blood filaments of the eyeball, The multi-column convolution neural network structure is designed: each layer of glioma samples is used as the input of each column and the number of layers of convolution and sub-sampling layers is increased. The Maxout activation function is used to replace the Sigmoid function which is often used in the traditional neural network. In the experiment, the multi-column structure and single-row structure are compared, and the results show the advantages of multi-column structure in glioma classification. In addition, the sample is optimized to further improve the classification accuracy. Finally, the multicolumn convolution neural network computation is visualized, and the working process of each layer is explained from the visual point of view.
【學(xué)位授予單位】：湖北工業(yè)大學(xué)
【學(xué)位級(jí)別】：碩士
【學(xué)位授予年份】：2017
【分類號(hào)】：TP391.41;TP183

【參考文獻(xiàn)】

相關(guān)期刊論文 前10條

1 李鳴;張鴻;;基于卷積神經(jīng)網(wǎng)絡(luò)迭代優(yōu)化的圖像分類算法[J];計(jì)算機(jī)工程與設(shè)計(jì);2017年01期

2 張舉森;陳菲;;一種基于模糊聚類的醫(yī)學(xué)圖像快速分割算法[J];西南科技大學(xué)學(xué)報(bào);2016年03期

3 張永梅;巴德凱;邢闊;;基于模糊閾值的自適應(yīng)圖像分割方法[J];計(jì)算機(jī)測(cè)量與控制;2016年04期

4 杜艷新;葛洪偉;肖志勇;;基于模糊連接度的近鄰傳播聚類圖像分割方法[J];計(jì)算機(jī)應(yīng)用;2014年11期

5 李海峰;李純果;;深度學(xué)習(xí)結(jié)構(gòu)和算法比較分析[J];河北大學(xué)學(xué)報(bào)(自然科學(xué)版);2012年05期

6 周光華;李岳峰;孟群;;醫(yī)學(xué)圖像處理技術(shù)與應(yīng)用分析[J];中國(guó)衛(wèi)生信息管理雜志;2011年06期

7 李清夢(mèng);聶生東;;神經(jīng)網(wǎng)絡(luò)技術(shù)及其在醫(yī)學(xué)圖像處理中的應(yīng)用[J];中國(guó)醫(yī)學(xué)影像技術(shù);2011年06期

8 朱海峰;陳雪功;;中醫(yī)診斷客觀化研究的現(xiàn)狀和展望[J];甘肅中醫(yī);2007年06期

9 英杰,李重錫,李梢,季梁,劉德麟,馬維婭;腦血管病患者舌象特征的提取與分析方法[J];北京中醫(yī)藥大學(xué)學(xué)報(bào);2005年04期

10 王今覺(jué);談“望目辨證”的中醫(yī)學(xué)理論基礎(chǔ)[J];中國(guó)中醫(yī)基礎(chǔ)醫(yī)學(xué)雜志;2005年05期

相關(guān)碩士學(xué)位論文 前3條

1 吳正文;卷積神經(jīng)網(wǎng)絡(luò)在圖像分類中的應(yīng)用研究[D];電子科技大學(xué);2015年

2 何鵬程;改進(jìn)的卷積神經(jīng)網(wǎng)絡(luò)模型及其應(yīng)用研究[D];大連理工大學(xué);2015年

3 羅蔓;結(jié)合MRI多模態(tài)信息與3D-CNNS特征提取的腦腫瘤分割研究[D];南方醫(yī)科大學(xué);2015年

，

本文編號(hào)：2072786