English  |  正體中文  |  简体中文  |  Items with full text/Total items : 28611/40649
Visitors : 646156      Online Users : 52
RC Version 4.0 © Powered By DSPACE, MIT. Enhanced by NTU Library IR team.
Scope Adv. Search
LoginUploadHelpAboutAdminister

Please use this identifier to cite or link to this item: http://ntour.ntou.edu.tw:8080/ir/handle/987654321/54115

Title: 多層自我組織映射圖應用於肺腫瘤病變檢測
Detection of Lung Tumors Based on Multilayer Self-Organization Maps
Authors: Pan, Chen
潘辰
Contributors: NTOU:Department of Electrical Engineering
國立臺灣海洋大學:電機工程學系
Keywords: 斷層掃描;肺腫瘤;自我組織映射;深度學習;病變檢測
CT;Lung Tumor;SOM;Deep-Learning;Lesion Detection
Date: 2019
Issue Date: 2020-07-09T03:02:05Z
Abstract:   電腦斷層掃描(Computed Tomography,簡稱CT)被廣泛應用於醫療診斷,現今仍由醫生憑經驗用肉眼判別病變區塊,但在長期判別病變區塊之下,身心感到疲累之下會影響判別;另一方面,近年來深度學習(Deep learning)崛起,深度學習被廣泛用用到許多領域,諸如自動駕駛、車牌辨識、語意識別等;在醫療上,本論文期望能將深度學習應用於醫學領域,藉由深度學習輔助醫生辨識肺腫瘤,希望能給予醫生較為客觀的判別,以減少醫生判斷的時間並降低誤判率。   本論文所使用的網路架構是結合三層包含自我組織映射(Self-organizing Maps,簡稱SOM)與深度神經網路(Deep Neural Networks,簡稱 DNN)的神經網路,建立一個自動檢測並標註肺腫瘤的神經架構;本研究會對我們所提出的SOM-DNN架構做進一步說明,並說明此架構是如何標註肺腫瘤影像。   本論文所提出的檢測模型目前能找出近9成的腫瘤區塊,特徵判別的部分能找到約8成,肺腫瘤的形狀辨識則是能到近9成左右;模型整體架構的F1 score可以達到8成以上。
Computed Tomography (CT) is widely used in medical diagnosis. Today, doctors discriminate lesions on gross. However, doctors may be stressed out under long-term discriminating lesions. On the other hand, deep learning (deep learning) has emerged in recent years, and deep learning has been widely used in many fields, such as autonomous driving, license plate recognition, semantic recognition, etc. In medical treatment, this paper hopes to applicate deep-learning method to the medical field. By deep learning to assist doctors in identifying lung tumors, it is hoped that doctors can be given more objective judgments to reduce their time for judging and reduce the rate of false positives. The network architecture in this paper is to combine a three-layer neural network containing Self-organizing Maps (SOM) and Deep Neural Networks (DNN) to establish an automatic detection and circle of lung tumors. The neural architecture; this study will further explain our proposed SOM-DNN architecture and explain how this architecture marks lung tumor areas. The detection model proposed in this paper can find close to 85% of the tumor blocks, the part of the feature identification can be found close to 80%, and the shape recognition of the lung tumor can reach about 90%. The F1 score of the model can reach more than 80%
URI: http://ethesys.lib.ntou.edu.tw/cgi-bin/gs32/gsweb.cgi?o=dstdcdr&s=G0010753030.id
http://ntour.ntou.edu.tw:8080/ir/handle/987654321/54115
Appears in Collections:[電機工程學系] 博碩士論文

Files in This Item:

File SizeFormat
index.html0KbHTML14View/Open


All items in NTOUR are protected by copyright, with all rights reserved.

 


著作權政策宣告: 本網站之內容為國立臺灣海洋大學所收錄之機構典藏,無償提供學術研究與公眾教育等公益性使用,請合理使用本網站之內容,以尊重著作權人之權益。
網站維護: 海大圖資處 圖書系統組
DSpace Software Copyright © 2002-2004  MIT &  Hewlett-Packard  /   Enhanced by   NTU Library IR team Copyright ©   - Feedback