臺灣博碩士論文加值系統

目前超音波檢查是診斷及追蹤肌肉纖維化的主要方法，對於經驗豐富的醫師而言有很好的準確度，但對初學者而言，較易忽略細微的變化，而影響病情的判斷，導致治療效果不佳。本論文嘗試利用影像分割於肌肉超音波影像進行纖維化程度的分析，首先提出一種混合型 (結合區域成長紋理方向與邊緣偵測)分割演算法，基於該分割結果再利用肌肉受傷評估演算法[6]來計算纖維化程度，希望達到既可符合人類視覺所觀察到肌肉紋理纖維的效果，又可提供一客觀量化指標來來描述肌肉纖維化程度。
實驗初期以11隻受傷老鼠的小腿腓腸肌，利用不同週期(W2~W4)分成三組來模擬人類肌肉纖維化的程度，每隻老鼠分別拍攝2張超音波影像與2~30張切片影像。本論文嘗試建立由超音波影像的特徵量測與切片纖維化百分比之關係，其實驗結果顯示若搭配傳統區域成長[22]作為影像前置處理，其效果無法彰顯不同週期的受傷程度，相較於此，本論文所提出的混合型分割演算法能區分不同週期受傷程度，且間接驗證超音波影像肌肉纖維化程度與組織切片結果有高度正相關(R=0.76、P=0.006)。

Ultrasonography is currently the mainstream method in clinically diagnosing and tracking the muscle fibrosis, it can provide rather accurate results for experienced physicians, but for a beginner it is likely susceptible to ignore subtle changes, which could adversely lead to erroneous judgment on the disease. This paper aims to applying image processing techniques to ultrasound image analysis of muscle fibrosis, firstly, a hybrid-type region growing algorithm is proposed to dissect directional regions of texture, and based on the segmentation results [6] to assess fibrosis degree, it is expected that not only muscle textures thus identified are in line with the human visual perception, but also an objective and quantitative indicator for evaluating the muscle fibrosis can be derived therefrom.
Experimental results on 11 mice calf gastrocnemius are shown to verify the performance of the presented approach, specifically, two ultrasound images (normal leg and injured leg) and 2-30 dyed tissue images are prepared respectively for each mouse at different weeks, and a fibrosis percentage (for the dyed tissue sliced images) and a muscle injury assessment score (for the ultrasound images) are derived to compute their correlation, and the positive Pearson correlation result justifies the effectiveness of the proposed approach.

摘要............................................I
Abstract.......................................II
目錄..........................................III
第一章 緒論.....................................1
1-1背景介紹.....................................1
1-2研究動機及目的...............................2
1-3論文架構.....................................4
第二章 肌肉纖維化................................5
2-1肌肉纖維化於超音波影像上的特徵.................5
2-2肌肉纖維化的程度判讀.........................6
第三章 研究方法.................................8
3-1影像來源.....................................8
3-2組織切片分析.................................9
3-2-1白平衡處理................................10
3-2-2纖維化百分比的計算........................10
3-3超音波影像分割..............................13
3-3-1影像前置處理..............................16
3-3-2混合型分割演算法..........................17
3-3-2-1肌肉紋理方向資訊選取與流程圖............18
3-3-2-2 Marr-Hildreth邊緣偵測..................20
3-3-2-3區域成長步驟與流程圖....................23
3-4肌肉纖維化的量測............................27
第四章 實驗結果與分析..........................31
4-1系統環境與動物實驗影像......................31
4-2實驗結果與分析..............................32
第五章 結論與未來方向..........................41
5-1 結論.......................................41
5-2 未來方向...................................41
參考文獻.......................................43

[1] E. McNally, Practical Musculoskeletal Ultrasound. Churchill Livingstone, Elsevier, Nov. 2004.
[2] S.W. Chou and M.K. Wong, “Outcome of Torticollis Children,” Journal of Rehabilitation Medicine Association, Dec. 1991.
[3] J. Menetrey, C. Kasemkijwattana, F. H. Fu, M. S. Moreland, and J.Huard, “Suturing versus immobilization of a muscle laceration: Amorphological and functional study in a mouse model,” Am. J. Sports. Med., vol. 27, no. 2, pp. 222-229, Mar. 1999.
[4] S. F. T. Tang, K. H. Hsu, A. M. K. Wong, C. C. Hsu, and C. H. Chang,“Longitudinal followup study of ultrasonography in congenitalmuscular torticollis,” Clin. Orthop. Relat. Res., vol. 403, pp. 179-185, Oct. 2002.
[5] 施宏潔《應用於骨骼肌超音波影像之紋理特徵分析》，國立海洋大學，碩士論文，2009。
[6] Jian-Jiun Ding, Yu-Hsiang Wang, Lee-Lin Hu, Wei-Lun Chao, and Yio-Wha Shau, “Muscle Injury Determination by Image Segmentation,” IEEE Visual Communications and Image Processing (VCIP), pp.1-4, 2011.
[7] Lin, J.N and M.-L.Chou, “Ultrasonographic study of the sternocleidomastoid muscle in the management of congenital muscular Torticollis,” Journal of Pediatric Surgery, pp.1648-1651, 1997.
[8] T.C. Hsu, C.L. Wang, M.K. Wong, K.H. Hsu, F.T. Tang, and H.T. Chen, “Correlation of clinical and ultrasonographic features incongenital muscular torticollis,” Archives of physical medicine and rehabilitation, vol. 80, pp. 637-641, 1999.
[9] Heckmatt, J.Z., S. Leeman, and V.Dubowitz, “Ultrasound imaging in the diagnosis of muscle disease,” The Journal of pediatrics, 101(5): p.656-660, 1982.
[10] S. Pillen, et al. “Skeletal muscle ultrasound: correlation between fibrous tissue and echo intensity,” Ultrasound Med. Biol., vol. 35, issue 3, pp.443-446, Mar. 2009.
[11] S. Pillen, R. R. Scholten, M. J. Zwarts, and A. Verrips, “Quan-titative skeletal muscle ultrasonography in children with sus-pected neuromuscular disease,” Muscle &; Nerve, vol. 27, issue 6, pp. 699-705, June 2003.
[12] 胡李琳《自動白平衡之軟硬體共同設計》，臺灣大學，碩士論文，2010。
[13] 林殷旭《自動白平衡之軟硬體共同設計》，臺北科技大學，碩士論文，2006。
[14] Tzan-Sheng Chiou, “Automatic White Balance for Digital Still Camera ,” MSc thesis, National Taiwan University, Department of Computer Science and Information Engineering, 2001.
[15] Trezzi, G.D.F.a.E., “Shades of gray and colour constancy, in Twelfth Color Imaging Conference: Color Science and Engineering Systems,” Technologies, and Applications. pp. 37-41, 2004.
[16] Su Wang, Yewei Zhang, Peng Deng, Fuqiang Zhou, “Fast automatic white balancing method by color histogram stretching,” Image and Signal Processing (CISP), 2011 4th International Congress on. pp.979-983, 2011.
[17] R. C. Gonzales and R. E. Woods, Digital Image Processing, 3rd ed., Prentice Hall, 2008.
[18] Marr, D. and Hildreth, E. “Theory of Edge Detection,” Proc. R.Soc.Lond., vol. B207, pp.187-217, 1980.
[19] C. S. Tseng, C. T. Lin, C. W. Lin, and J. H. Wang, “Perceptual Edges Preservation Conformal to Human Vision Perception,” The 13th IEEE International Conf. on Information Reuse and Integration, Aug. 2012.
[20] A.R. Kavitha, C. Chellamuthu, K. Rupa, “An efficient approach for brain tumour detection based on modified region growing and neural network in MRI images,” IEEE International Conference on Computing, Electronics and Electrical Technologies (ICCEET), pp3-4, 2012.
[21] B.V. Dasarathy and E.B. Holder, “Image characterizations based on joint gray level-run length distributions,” Pattern Recogn. Lett., vol. 12, pp. 497-502, 1991.
[22] J. J. Ding, C. J. Kuo, and W. C. Hong, “An efficient image segmentation technique by fast scanning and adaptive merging,” CVGIP, Aug. 2009.