Please use this identifier to cite or link to this item:
DC FieldValueLanguage
dc.contributor.authorYou-Kuan Chaoen_US
dc.description.abstract乳癌為女性十大死因之一,而乳腺的密度高低和乳癌的發生率息息相關,近年來磁振造影影像(Magnetic Resonance Imaging)已漸漸取代傳統的X光和超音波檢測作為乳癌檢查的工具,磁振造影影像不但沒有放射線的問題且有較優的影像解析度及對比度。因此,本研究提出一個方法,藉由使用IVIM-MRI影像分類乳房組織。 本研究以MRI之T1、T2、PD等影像作為輸入,為求可以準確偵測出乳腺位置,我們結合支持向量機和限制能量最小化法來分類出乳腺和脂肪,並且把乳腺及脂肪在T1、T2中的位置對應到IVIM影像中,再以IVIM分析出屬於脂肪和乳腺所對應到的五個參數(D、D*、P、PF、slope),計算該係數包含外顯擴散係數(Apparent Diffusion Coefficient)、不同b 值加權影像後之訊號衰減變化程度(Slope)、純擴散(pure-diffusion)、血液灌注(perfusion-related diffusion)及灌注因子(perfusion fraction),累計參數統計後將定義乳腺和脂肪的臨界值,以達到不需要使用T1、T2…等影像,直接使用IVIM就能做出乳腺及脂肪分析。zh_TW
dc.description.abstractBreast cancer is one of the top ten cause of death among women in the world. Breast density is positively correlated with breast cancer. In recent years, Magnetic Resonance Imaging has gradually taken the place of traditional X-ray and ultrasound detection as a tool for breast cancer screening. Using the MR image not only has no radiation problems, but also has excellent image resolution. Therefore, we developed a method for classifying breast tissue using IVIM MR images. MR images such as T1, T2, and PD were used as inputs for glandular detection. In order to closely examine the location of the glandular, we used constrained energy minimization (CEM) method to detect glandular and fat and used them as correct positions. After getting the positions of glandular and fat, we use them to map on the IVIM MR images, and then analyzing the five parameters (D, D*, ADC, PF, and slope). Parameters is including Apparent Diffusion Coefficient, different b values weighted on different signal attenuation, perfusion-related diffusion and perfusion fraction. The accumulation of parameter statistics will define the critical values of glandular and fat. Finally, when we only input MR IVIM MR images, breast tissues can be classified without using T1, T2 or PD images.en_US
dc.description.tableofcontents致謝 i 中文摘要 ii Abstract iii Contents iv List of Figures vi List of Table x CHAPTER 1 Introduction 1 CHAPTER 2 Background 3 2.1 Magnetic Resonance Imaging 3 2.2 Diffusion-Weighted Imaging (DWI) 3 2.3 Intravoxel incoherent motion (IVIM) 4 2.4 Band expansion process (BEP) 6 2.5 Constrained energy minimization approach 7 2.6 Histogram Analysis 9 2.7 Decision Tree 10 CHAPTER 3 Methods 11 3.1 Experimental Material 11 3.2 Image pre-processing 12 3.3 Glandular and fat tissues detection 14 3.3.1 Band expansion process 14 3.3.2 Constrained energy minimization approach 15 3.4 Quantitative breast tissue 16 3.5 Using Decision Tree to Classify Breast Tissue 17 CHAPTER 4 Experiment Results 18 4.1 Real MRI images cases 18 4.2 glandular tissue and fat tissue 19 4.3 Result of glandular tissue histogram analysis 40 4.4 Result of fat tissue histogram analysis 45 4.5 Comparison of glandular tissue and fat tissue 53 CHAPTER 5 Conclusion 56 Reference 57zh_TW
dc.subjectmagnetic resonance imagingen_US
dc.subjectintra-voxel incoherent motionen_US
dc.subjectdiffusion weight imagingen_US
dc.subjectconstrained energy minimizationen_US
dc.titleQuantitative Measurement of Breast Tissue on Intravoxel Incoherent Motion(IVIM) MR Imageen_US
dc.typethesis and dissertationen_US
Appears in Collections:通訊工程研究所


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.