本研究開發一套自動分割淋巴結於電腦斷層掃描影像，並自動計算出短徑、長徑、長短徑比值與體積資訊，提供醫生利用淋巴結評估腫瘤時能有更多及更客觀的參考資訊，最後對淋巴結進行體積表面重建，提供醫生了解整顆淋巴結的形狀結構，以利在診斷時能有更多資訊做為參考。本研究可分為兩部分，第一部分提出一套影像處理系統對電腦斷層掃描中的淋巴結影像進行圈選及計算：首先利用中央加權中值濾波器影像前處理，排除影像受到雜訊及外界的干擾，並保留影像細節紋理不受破壞。接著使用最大熵值法配合適應性K-means完成影像分群。透過分群可以使每次分割及各個值的量測結果不受到初始點位置影響，克服放射師在圈選及量測時造成的結果不一致性。但淋巴結部分可能會與鄰近器官組織相連，因此在分群之後使用分水嶺演算法針對相鄰部分進行修正。輪廓提取部分考慮到速度及準確性選用Canny邊緣檢測，且根據特徵條件比較前後張影像是否還有淋巴結的部分，最後利用邊緣檢測結果及特徵判斷條件完成淋巴結的體積表面重建。
第二部分為醫學指標分析，目前淋巴結對於腫瘤的侵襲評估判斷以短徑10mm作為依據，本研究欲以其他指標如長徑、比值、體積等條件比較目前短徑10mm對於腫瘤的侵襲性評估是否更具準確性，評估結果由比較ROC曲線下面積判斷，經實驗發現體積對於腫瘤侵襲性評估的ROC曲線下面積為0.90優於短徑的ROC曲線下面積0.82，說明了淋巴結體積與腫瘤侵襲具有高度相關性，而ROC最佳分界點的敏感度與特異度分別為90.0%及85.6%。
本研究利用適應性分群演算法克服淋巴結分割結果受到初始點的影響，並使分割結果具有一致性，提供醫生在評估淋巴結時有更客觀的參考資訊，最後建立淋巴結對於腫瘤侵襲性之指標。

This study developed automatic segmentation of lymph node in CT image, and the short diameter, long diameter, long-short diameter ratio and volume information were calculated automatically, so that the doctors have more objective reference information when using lymph node to evaluate the tumor. Finally, the volume surface of lymph node was reconstructed for the doctors to learn about the shape structure of the whole lymph node, with more reference information for diagnosis. This study was comprised of two parts, Part 1 proposed an image processing system to circle and calculate the lymph node image in CT. First, the center-weighted median filter image preprocessing was implemented to eliminate the noises and external interference from the image and to protect the image detail texture from damage. Afterwards, the maximum entropy method is combined with adaptive K-means to complete the image clustering. The clustering made the measurement results of each segmentation free from the initial point position influence, overcoming the result inconsistency of radiologist during circling and measurement. However, the lymph node portion may be connected to adjacent organ tissues, so the watershed algorithm is used after clustering to correct adjacent portions. For contour extraction, the Canny edge detection is selected considering the speed and accuracy, and the former and latter images were compared according to characteristic condition to check whether there were lymph nodes or not. Finally, the volume surface reconstruction of lymph node was completed by using the edge detection result and feature judgment conclusion.
Part 2 was medical indicator analysis. Currently, the lymph node for tumor invasion evaluation was based on short diameter 10 mm, this study would use other indexes, such as long diameter, ratio and volume conditions, to compare whether the present short diameter 10mm was more accurate for tumor invasion evaluation. The evaluation result was judged by comparing the area under ROC curve. The experiment showed that the volume for the area under ROC curve 0.90 of tumor invasion evaluation is better than the area under ROC curve 0.82 of short diameter, meaning the lymph node volume is highly correlated with tumor invasion, and the sensitivity and specificity of ROC optimum cut-off point are 90.0% and 85.6% respectively.
This study used adaptive clustering algorithm to overcome the effect of initial point on the lymph node segmentation result, and if the segmentation result was consistent, the doctors were provided with more objective reference information when evaluating the lymph node. Finally, the indexes of lymph node for tumor invasiveness were established.

1. Doi, K., Current status and future potential of computer-aided diagnosis in medical imaging. The British journal of radiology, 2005. 78(suppl_1): p. s3-s19.
2. Bruneton, J.N., et al., Axillary lymph node metastases in breast cancer: preoperative detection with US. Radiology, 1986. 158(2): p. 325-326.
3. Pamilo, M., M. Soiva, and E.-M. Lavast, Real‐time ultrasound, axillary mammography, and clinical examination in the detection of axillary lymph node metastases in breast cancer patients. Journal of ultrasound in medicine, 1989. 8(3): p. 115-120.
4. Prenzel, K.L., et al., Lymph node size and metastatic infiltration in non-small cell lung cancer. Chest, 2003. 123(2): p. 463-467.
5. Cho, J.-K., et al., Significance of lymph node metastasis in cancer dissemination of head and neck cancer. Translational oncology, 2015. 8(2): p. 119-125.
6. Organization, W.H., WHO handbook for reporting results of cancer treatment. 1979.
7. Eisenhauer, E., et al., New response evaluation criteria in solid tumours: revised RECIST guideline (version 1.1). European journal of cancer, 2009. 45(2): p. 228-247.
8. Schwartz, L., et al., Evaluation of lymph nodes with RECIST 1.1. European journal of cancer, 2009. 45(2): p. 261-267.
9. Beichel, R.R. and Y. Wang, Computer‐aided lymph node segmentation in volumetric CT data. Medical physics, 2012. 39(9): p. 5419-5428.
10. Honea, D.M. and W.E. Snyder. Three-dimensional active surface approach to lymph node segmentation. in Medical Imaging'99. 1999. International Society for Optics and Photonics.
11. Yan, J., et al., Lymph node segmentation from CT images using fast marching method. Computerized Medical Imaging and Graphics, 2004. 28(1): p. 33-38.
12. Yan, J., et al., Marker‐controlled watershed for lymphoma segmentation in sequential CT images. Medical Physics, 2006. 33(7): p. 2452-2460.
13. Dornheim, J., et al., Segmentation of neck lymph nodes in CT datasets with stable 3D mass-spring models: Segmentation of neck lymph nodes. Academic Radiology, 2007. 14(11): p. 1389-1399.
14. Steger, S., D. Ebert, and M. Erdt. Lymph node segmentation in CT slices using dynamic programming. in Biomedical Imaging: From Nano to Macro, 2011 IEEE International Symposium on. 2011. IEEE.
15. Yu, P. and C.L. Poh, Region-based snake with edge constraint for segmentation of lymph nodes on CT images. Computers in biology and medicine, 2015. 60: p. 86-91.
16. Hartigan, J.A. and M.A. Wong, Algorithm AS 136: A k-means clustering algorithm. Journal of the Royal Statistical Society. Series C (Applied Statistics), 1979. 28(1): p. 100-108.
17. Bezdek, J.C., R. Ehrlich, and W. Full, FCM: The fuzzy c-means clustering algorithm. Computers & Geosciences, 1984. 10(2-3): p. 191-203.
18. Ghosh, S. and S.K. Dubey, Comparative analysis of k-means and fuzzy c-means algorithms. International Journal of Advanced Computer Science and Applications, 2013. 4(4): p. 35-39.
19. Chen, C.W., J. Luo, and K.J. Parker, Image segmentation via adaptive K-mean clustering and knowledge-based morphological operations with biomedical applications. IEEE transactions on image processing, 1998. 7(12): p. 1673-1683.
20. Ahmed, M.N., et al., A modified fuzzy c-means algorithm for bias field estimation and segmentation of MRI data. IEEE transactions on medical imaging, 2002. 21(3): p. 193-199.
21. Li, B.N., et al., Integrating spatial fuzzy clustering with level set methods for automated medical image segmentation. Computers in biology and medicine, 2011. 41(1): p. 1-10.
22. Xu, Z., J. Chen, and J. Wu, Clustering algorithm for intuitionistic fuzzy sets. Information Sciences, 2008. 178(19): p. 3775-3790.
23. Kurland, B.F., et al., Promise and pitfalls of quantitative imaging in oncology clinical trials. Magnetic resonance imaging, 2012. 30(9): p. 1301-1312.
24. Harisinghani, M.G. and R. Weissleder, Sensitive, noninvasive detection of lymph node metastases. PLoS medicine, 2004. 1(3): p. e66.
25. Schnyder, P. and G. Gamsu, CT of the pretracheal retrocaval space. American Journal of Roentgenology, 1981. 136(2): p. 303-308.
26. Hanahan, D. and R.A. Weinberg, The hallmarks of cancer. cell, 2000. 100(1): p. 57-70.
27. Urban, D., et al., Lymph Node Ratio May Predict the Benefit of Postoperative Radiotherapy in Non–Small-Cell Lung Cancer. Journal of Thoracic Oncology, 2013. 8(7): p. 940-946.
28. Travis, W.D., et al., International association for the study of lung cancer/american thoracic society/european respiratory society international multidisciplinary classification of lung adenocarcinoma. Journal of thoracic oncology, 2011. 6(2): p. 244-285.
29. Erasmus, J.J., et al., Interobserver and intraobserver variability in measurement of non–small-cell carcinoma lung lesions: Implications for assessment of tumor response. Journal of Clinical Oncology, 2003. 21(13): p. 2574-2582.
30. Schwartz, L.H., et al., Variability in response assessment in solid tumors. Clinical cancer research, 2003. 9(12): p. 4318-4323.
31. Glazer, G.M., et al., Normal mediastinal lymph nodes: number and size according to American Thoracic Society mapping. American Journal of Roentgenology, 1985. 144(2): p. 261-265.
32. Matsuoka, H., et al., Morphological characteristics of lateral pelvic lymph nodes in rectal carcinoma. Langenbeck's Archives of Surgery, 2007. 392(5): p. 543-547.
33. Saito, T., et al., Accuracy of multidetector-row CT in diagnosing lymph node metastasis in patients with gastric cancer. European radiology, 2015. 25(2): p. 368.
34. Buerke, B., et al., Measurement accuracy and reproducibility of semiautomated metric and volumetric lymph node analysis in MDCT. American Journal of Roentgenology, 2010. 195(4): p. 979-985.
35. Beyer, F., et al., Prediction of lymph node metastases in NSCLC. Nuklearmedizin, 2010. 49(1): p. 41.
36. Siegel, R.L., et al., Colorectal cancer statistics, 2017. CA: a cancer journal for clinicians, 2017. 67(3): p. 177-193.
37. Siegel, R.L., K.D. Miller, and A. Jemal, Cancer statistics, 2016. CA: a cancer journal for clinicians, 2016. 66(1): p. 7-30.
38. Surgeons, T.S.o.T. The Patient Guide to Heart, Lung, and Esophageal Surgery. Available from: http://ctsurgerypatients.org/lung-esophageal-and-other-chest-diseases/mediastinal-tumors.
39. Library, C. Lungs PNG Transparent Images. Available from: http://clipart-library.com/lungs-png.html.
40. Rusch, V.W., et al., The IASLC lung cancer staging project: a proposal for a new international lymph node map in the forthcoming seventh edition of the TNM classification for lung cancer. Journal of Thoracic Oncology, 2009. 4(5): p. 568-577.
41. Edge, S.B. and C.C. Compton, The American Joint Committee on Cancer: the 7th edition of the AJCC cancer staging manual and the future of TNM. Annals of surgical oncology, 2010. 17(6): p. 1471-1474.
42. Wang, Z. and D. Zhang, Progressive switching median filter for the removal of impulse noise from highly corrupted images. IEEE Transactions on Circuits and Systems II: Analog and Digital Signal Processing, 1999. 46(1): p. 78-80.
43. Gajdhane, M.V.A. and L. Deshpande, Detection of Lung Cancer Nodule on Computed Tomography Images by Using Image Processing. Communications, 2014. 3(7).
44. Kitasaka, T., et al., Automated extraction of lymph nodes from 3-D abdominal CT images using 3-D minimum directional difference filter. Medical Image Computing and Computer-Assisted Intervention–MICCAI 2007, 2007: p. 336-343.
45. Punarselvam, E. and R. Parthasarathy, Segmentation of CT scan lumbar spine image using median filter and canny edge detection algorithm. International Journal on Computer Science and Engineering, 2013. 5(9): p. 806.
46. Otsu, N., A threshold selection method from gray-level histograms. IEEE transactions on systems, man, and cybernetics, 1979. 9(1): p. 62-66.
47. Bhatia, S.K. Adaptive K-Means Clustering. in FLAIRS Conference. 2004.
48. Meyer, F., Topographic distance and watershed lines. Signal processing, 1994. 38(1): p. 113-125.
49. Maurer, C.R., R. Qi, and V. Raghavan, A linear time algorithm for computing exact Euclidean distance transforms of binary images in arbitrary dimensions. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2003. 25(2): p. 265-270.
50. Canny, J., A computational approach to edge detection. IEEE Transactions on pattern analysis and machine intelligence, 1986(6): p. 679-698.
51. Hanley, J.A. and B.J. McNeil, The meaning and use of the area under a receiver operating characteristic (ROC) curve. Radiology, 1982. 143(1): p. 29-36.
52. Youden, W.J., Index for rating diagnostic tests. Cancer, 1950. 3(1): p. 32-35.