臺灣博碩士論文加值系統

磁振造影的影像分割有許多臨床上的實際用途：輔助醫生的診斷及治療、使流行病學統計容易進行、或是用於各類醫學研究的資料分析。然而多數時候，影像分割仰賴人工操作，費時費力且難以標準化。肌肉是人體重要組成成份，之前對身體其他部份，如腦部的自動影像分割，已有不少成果，然而尚無肌肉磁振影像的自動分割研究。
本論文的目的是發展一套自動分割人體下肢磁振影像上不同功能肌肉群之工具，可用於日漸增加的肌肉組織相關的研究，例如追蹤肌萎縮患者之病情變化或是不同型態運動員的肌肉運用與表現。由於不同身體部位的肌肉解剖型態及其週邊結構差異很大，為了方便及因應本實驗室其他研究之需求，本研究選擇以骨盆腔至小腿之肌肉為研究對象。又因為不同肌肉的磁振影像特性極為相似，難以自動區分單一之肌肉，因此本研究僅以自動分割出不同功能肌肉群為研究標的。
論文中將骨盆至小腿的肌肉依相鄰及相似為原則，由上而下分成數段處理。藉由每個區段的解剖知識輔助，例如皮下脂肪、骨盆的腸骨、大小腿骨及肌肉間筋膜及膝蓋的軟骨等在磁振影像的訊號與肌肉相差較大，可用為分界。再利用數學形態學中的侵蝕、膨脹、斷開、閉合等方法，及其他影像處理技巧如區域成長和區域填充等演算法，將八種不同功能的肌肉群一一分類出來。此自動分類再與物理治療師以手動方式所做出的分類做比較。結果顯示肌肉自動分類的總合平均絕對誤差(百分比誤差) ± 標準差為1736.5 (13.5%) ± 1071.8 ml。其中以膝伸肌的490.2 (14.0%) ± 371.8 ml為最大絕對誤差，而以踝屈肌的41.3 (7.7%) ± 11.6 ml為最小絕對誤差。相對誤差則以髖屈肌的242.8 (24.4%) ± 140.2為最大，踝伸肌的153.1 (6.7%) ± 94.3為最小。
本論文使用型態學之方法輔以解剖知識，成功發展出一套可以自動分類出人體下肢磁振影像不同功能的肌肉群之工具，未來可以再改進其準確度並應用於身體的其他部位，如臟器脂肪自動定量，應可協助醫生的診斷及治療或是幫助相關的研究。

Segmentation of magnetic resonance (MR) images has numerous clinical applications: Being an auxiliary tool in diagnosis and treatment, making epidemiological statistics easier to carry on, being an important step in data analysis of all kinds of medical researches. However, most of the segmentation has been done manually or at most semi-automatically. The process is time- and energy-consuming and difficult to be standardized. Automatic segmentation algorithms of MR images for some of the body parts, such as the brain, have been developed with success. The muscular tissue is a major component of the human body, however, to our knowledge, no similar studies had been done on automatic muscle segmentation on MR images.
The goal of this study is to develop an automatic segmentation scheme to correctly assign different functional muscle groups on MR images of the human lower extremities. We speculated that the results could be used in increasing muscle-tissue-related researches, such as the monitoring of muscle volume change over the time for the victims of muscular dystrophy diseases and investigation of the use and the performance of different muscles in athletes of various sport types. The grouping and the surrounding anatomic structures are quite different for muscles located at different body parts. For convenience and to support another research of our labs, we chose to target at lower extremities, muscles from pelvis to ankle, as the object of the study. Since the MR signals for all muscles are more or less similar, instead of singling out an individual muscle, our study focused on automatic classification of functional muscle groups.
In the thesis, the lower extremities were first divided into several longitudinal anatomic segments based on the principles of proximity and anatomic similarity. Since the MR signals are quite different among bones, fascia, fat, and muscles, subcutaneous fat, ilium of the pelvis, femur, tibia, fibula, meniscus, and the muscular fascia are available as boundaries. Equipped by this knowledge, we applied mathematical morphological operations such as erosion, dilation, open, and close, and other image processing techniques such as regional growing and filling to designate muscle areas on an MR image as one of the eight muscle functional groups. The results of the automatic segmentation were then compared against the classification manually made by a physical therapist. We found that the average total absolute error (relative error) ± the standard deviation of our automatic segmentation is 1736.5 (13.5%) ± 1071.8 ml, with 490.2 (14.0%) ± 371.8 ml of the knee extensors as the largest absolute error and 41.3 (7.7%) ± 11.6 ml of the ankle flexors as the smallest. On the other hand, the largest relative error is hip flexors’ 242.8 (24.4%) ± 140.2 ml and the smallest the ankle extensors’ 153.1 (6.7%) ± 94.3 ml.
The study presents that the combined mathematical morphology and human anatomy knowledge approach successfully divided muscles of lower extremity MR images into meaningful functional groups without human intervention. In the future, the accuracy of this method could be further improved by more sophisticated revision such as MR-atlas registration. Applications on other body parts and tissues such as abdominal visceral fat are under investigation. We expect the results of this and related studies to be helpful in body-composition-related researches and perhaps also in clinical diagnosis and treatment.

目錄
中文摘要 I
英文摘要 II
致謝 IV
目錄 V
圖目錄 VIII
表目錄 XI

第一章、緒論 1
1.1前言 1
1.2研究動機與目的 2
1.3論文架構 3
第二章、磁振造影之原理與應用 4
2.1磁振造影成像原理 4
2.2磁振造影硬體結構 7
2.2.1 磁鐵系統 7
2.2.2 射頻系統 7
2.2.3 電腦圖像重建系統 7
2.3 磁振造影在醫學上的應用 8
2.3.1 原理概述 8
2.3.2 磁振造影之優點 8
2.3.3 磁振造影之缺點及可能的危害 8
第三章、影像形態學之原理與應用 10
3.1原理概述 10
3.2膨脹與侵蝕演算法 13
3.2.1 原理 13
3.2.2 程式驗證 14
3.3斷開與閉合演算法 19
3.3.1 原理 19
3.3.2 程式驗證 20
3.4 區域成長 24
3.4.1 原理 24
3.4.2 程式驗證 24
3.5 區域填補 29
3.5.1 原理 29
3.5.2 程式驗證 29
第四章、研究資訊與流程 30
4.1研究資訊 30
4.1.1硬體資訊 30
4.1.2軟體程式工具 30
4.1.3肌肉圖譜資訊 31
4.1.4磁振影像資訊 32
4.1.5肌肉群分類表 33
4.2研究流程 34
4.2.1肌肉影像擷取 34
4.2.2肌肉影像分區 39
4.2.3肌肉影像修飾 41
4.2.4肌肉影像分類 48
第五章、研究結果與討論 54
5.1肌肉影像擷取結果 54
5.2肌肉影像分區結果 55
5.3肌肉影像修飾結果 56
5.4肌肉影像分類結果 58
5.5討論 63
第六章、結論與未來展望 64
第七章、參考文獻 65


圖目錄
圖2.1、外加磁場後原子核排列示意圖 4
圖3.1、邏輯運算示意圖 11
圖3.2、集合基本概念 12
圖3.3、膨脹示意圖 13
圖 3.4、侵蝕示意圖 14
圖3.5、幾何測試影像 15
圖3.6、膨脹結果 15
圖3.7、膨脹結果與原圖之差異 16
圖3.8、侵蝕結果 17
圖3.9、侵蝕結果與原圖之差異 18
圖3.10、斷開示意圖 19
圖3.11、閉合示意圖 20
圖3.12、斷開結果 21
圖3.13、斷開結果與原圖之差異 21
圖3.14、閉合結果 22
圖3.15、閉合結果與原圖之差異 23
圖3.16、區域成長結果 25
圖3.17、區域成長結果與原圖之差異 25
圖3.18、使用3X3結構元素區域成長結果 26
圖3.19、使用3X3結構元素區域成長結果與原圖之差異 27
圖3.20、使用7X7結構元素區域成長結果 28
圖3.21、使用7X7結構元素區域成長結果與原圖之差異 28
圖3.22、區域填補結果與原圖之差異 29
圖4.1、SIGNA EXCITE 1.5T MRI系統 30
圖4.2、ＭICROSOFT VISUAL STUDIO C++軟體工具 31
圖4.3、肌肉圖譜 32
圖4.4、肌肉分類流程圖 34
圖4.5、肌肉擷取流程圖 34
圖4.6、直接取閥值之結果 35
圖4.7、製作遮罩流程圖 36
圖4.8、取閥值後的結果 36
圖4.9、區域填充後的結果 37
圖4.10、OPEN和CLOSE前後比較 38
圖4.11、CLOSE後與合併後之比較 38
圖4.12、分區流程圖 39
圖4.13、骨盆區與腿部之比較 40
圖4.14、骨盆區與腿部Ｘ軸肌肉累計圖 40
圖4.15、肌肉累計曲線圖 41
圖4.16、肌肉修飾流程圖 42
圖4.17、骨盆區之非目標肌肉 42
圖4.18、骨盆區肌肉修飾流程圖 43
圖4.19、大腿上半部與骨盆區之比較 44
圖4.20、大腿區之非目標肌肉 44
圖4.21、大腿區肌肉修飾流程圖 45
圖4.22、膝蓋區之非目標肌肉 45
圖4.23、膝蓋區肌肉修飾流程圖 46
圖4.24、膝蓋區之骨骼間隙誤取部分 47
圖4.25、小腿區肌肉修飾流程圖 47
圖4.26、肌肉影像分類流程圖 46
圖4.27、大腿上半部和大腿區分類之差異 48
圖4.28、腸骨示意圖 49
圖4.29、髖外展肌和髖伸肌分類圖 49
圖4.30、骨盆區分類流程圖 50
圖4.31、大腿分類示意圖 50
圖4.32、大腿區分類流程圖 51
圖4.33、膝蓋分類示意圖 52
圖4.34、膝蓋區分類流程圖 52
圖4.35、小腿區分類示意圖 53
圖4.36、小腿區分類流程圖 53
圖5.1、粗略肌肉二值化影像 54
圖5.2、修飾後骨盆區肌肉二值化影像 56
圖5.3、修飾後大腿區肌肉二值化影像 56
圖5.4、修飾後膝蓋區肌肉二值化影像 57
圖5.5、修飾後小腿區肌肉二值化影像 57
圖5.6、各肌肉群肌肉PIXEL數累計圖 58
圖5.7、肌肉群累計誤差圖 59
圖5.8、肌肉體積累計圖 60
圖5.9、誤差肌肉體積累計圖 60
圖5.10、肌肉群平均體積統計圖 61
圖5.11、肌肉群平均誤差體積統計圖 62
圖5.12、各別肌肉平均誤差統計圖 62


表目錄
表2.1、人體常見原子核及其Ｉ值 4
表3.1、邏輯運算表 10
表4.1、磁振造影資訊表 32
表4.2、肌肉群分類表 33
表5.1、肌肉分區表 55
表5.2、肌肉群累計之肌肉數（PIXEL） 58
表5.3、肌肉群累計誤差表 58
表5.4、人體取樣表 59
表5.5、各肌肉群體積表 59
表5.6、肌肉誤差體積表 60
表5.7、平均肌肉體積和標準差 61
表5.8、肌肉平均誤差與標準差 61
表5.9、各別肌肉誤差百分比 62

[1] Wing P. Chan and Gin-Chung Liu , “MR Imaging of Primary Skeletal Muscle Diseases in Children” , AJR , 179,2002
[2] Suh-Jun Tai, Ren-Shyan Liu, Ya-Chen Kuo, Chi-Yang Hsu, and Chi-Hsien Chen ,” Glucose Uptake Patterns in Exercised Skeletal Muscles of Elite Male Long-Distance and Short-Distance Runners” , Chinese Journal of Physiology 53(2) , 2010
[3] Peter A. Rinck , “Magnetic Resonance in Medicine”
[4] Mark A. Brown and Richard C. Semelka , “MRI Basic Principles and Applications”
[5] 傅傑青, “核磁共振——獲得諾貝爾獎次數最多的一個科學專題”,自然雜誌,(06) 357-261 ,2003
[6] 別業廣、呂樺 ,”再談核磁共振在醫學方面的應用”,物理與工程, (02):34- 61,2004
[7] 金永君、艾延寶,”核磁共振技術及應用”,物理與工程,(01): 47-48,2002
[8] 黃衛華,”走近核磁共振”,醫藥與保健, (03):15, 2004
[9] 葉朝輝,”磁共振成像新進展”,物理,(01):12-17,2004
[10] 田建廣、劉買利、夏照帆、葉朝輝,”磁共振成像的安全性”,波譜學雜誌,(06):505-511,2002
[11] Rafael C. Gonzalez、Richard E, Woods ,”Digital Image Processing” , PEARSON
[12] Milan Sonka , Vaclav Hlavac , Roger Boyle, “ Image Processing Analysis, and Machine Vision Third Edition ”,THOMSON
[13] Canny, J.F.,”A Computational Approach To Edge Detection” ,IEEE Tans. Pattern Analysis and Machine Intelligence,8:679-714,1986
[14] Bleau A., Guise J. d, and LeBlanc R.A new set of fast algorithms for mathematical morphology II. Identification of topographic features on grayscale image. CVGIP: Image Understanding, 56(2): 210-229,1992
[15] Blum H. A transformation for extracting new descriptors of shape. In Wathen-Dunn W., editor, Proceeding of the Symposium on Models for the Perception of Speech and Visual Form, pages 362-380, Cambridge, MA,1967.MIT Press.
[16] Dougherty E.R. An Introduction to Mathematical Morphology Processing. SPIE Press, Belling-ham, WA, 1992
[17] Giardina C. R. and Dougherty E. R. Morphological Methods in Image and Signal Processing. Prentice-Hall, Englewood Cliffs, NJ,1998
[18] Haralick R. M., Stenberg S. R., and Zhuang X. Image analysis using mathematical morphology . IEEE Transaction on Pattern Analysis and Machine Intelligence, 9(4): 532-550, 1987
[19] Haralick R. M., Katz P. L., and Dougherty E. R. Model-based morphology: The opening spectrum. Graphical Models and Image Processing, 57(1): 1-12, 1995
[20] Heijmans H. J. Morphological Image Operators. Academic Press, Boston, 1994
[21] Maragos P. and Ziff R. Threshold superposition in morphological image analysis. IEEE Transactions on Pattern Analysis and Machine Intelligence, 12(5), 1990
[22] Roerdink J. B. and Heijmans H. J. Mathematical morphology for structures without translation symmetry. Signal Processing, 15(3): 271-277, 1988
[23] Serra J. Image Analysis and Mathematical Morphology. Academic Press, London, 1982
[24] Serra J. Morphological Optics. Journal of Microscopy, 145(1): 1-22, 1987
[25] Soille P. Morphological Image Analysis. Springer-Verlag, Berlin, 2 edition, 2003
[26] Vincent L. Fast opening function and morphological granulometries. In Proceedings Image Algebra and Morphological Image Processing V, page 253-267, San Diego, CA ,1994.SPIE
[27] 劉東華、李顯耀、孫朝暉,”核磁共振成像”,”大學物理”，1997,（10）:36-39, 29
[28] 阮萍,”核磁共振成像及其醫學應用”,”廣西物理”,1999,（02）:50-53, 28
[29] Lauterbur P C Nature, 1973, 242:190
[30] http://zh.wikipedia.org/zh-tw/MRI