The primary objective of this chapter is to develop a medical system to facilitate the diagnosis/evaluation of the treated region with LT. In segmentation module, the proposed knowledge-based random walks segmentation method can be applied to segment the livers and tumors from CT abdominal image. In registration mod-ule, we extended a classical intensity-based non-rigid registration algorithm using
an anatomical structure term to constrain the deformation. It can automatically register the segmented treated region after LT to the segmented tumor before LT.
From the field of views on application, our system can be regarded as an semi-automatic 3D imaging fusion technology that assess HCC of the treated region before and after LT. Fusion images enable easier understanding of the relationship between the tumor and ablation region, thus, helping evaluation of the therapeutic efficiency of HCC. This developed system can predict the local recurrence after LT for HCC. Moreover, the medical specialists gave the assessments on this system and pointed out that this work can be used in the clinical applications in the future.
Bibliography
[1] R. Inokuchi, T. Seki, K. Ikeda, R. Kawamura, K. Okazaki, A. Komemushi, N.
Omura, N. Tanigawa, S. Sawada: Assessment of treated margin after locore-gional therapy for hepatocellular carcinoma suing 3D fusion imaging. Journal of Kanzo (in Japan), 53(5)(2012) 298-301.
[2] C.H. Dong, T. Seki, R. Inoguchi, C.L. Lin, X.H. Han, Y.W. Chen: Nonrigid reg-istration for evaluating locoregional therapy of hepatocellular carcinoma. IEEE International Conference on Biomedical Engineering and Informatics (BMEI), (2013) 811-816.
[3] A. Sotiras, C. Davatzikos, N. Paragios: Deformable medical image registration:
A survey. IEEE Transactions on Medical Imaging, 32(7)(2013) 1253-1190.
[4] D. Rueckert, L.I. Sonoda, C. Hayes, D.L.G. Hill, M.O. Leach, D.J. Hawkes:
Nonrigid registration using free-form deformations application to breast MR images. IEEE Transactions on Medical Imaging, 18(8)(1999) 712-721.
[5] W.R. Crum, T. Hartkens, D.L. Hill: Non-rigid image registration: theory and practice. Journal of British Journal of Radiology, 77(2)(2004) S140-S153.
[6] D.L.G. Hill, P.G. Batchelor, M. Holden, D.J. Hawkes: Medical image registra-tion. Journal of Physics in Medicine and Biology, 46(3)(2001) R1-R45.
[7] J. Maintz, M.A. Viergever: A survey of medical image registration. Journal of Medical Image Analysis, 2(1)(1998) 1-36.
[8] D. Rueckert, J.A. Schnabel: Medical image registration. Journal of Biomedical Image Processing, (2011) 131-154.
[9] B. Zitova, J. Flusser: Image registration methods: a survey. Journal of Image and Vision Computing, 21(11)(2003) 977-1000.
[10] F. Maes, A. Collignon, D. Vandermeulen, G. Marchal, P. Suetens: Mul-timodality image registration by maximization of mutual information. IEEE Transactions on Medical Imaging, 16(2)(1997) 187-198.
[11] D. Mattes, D.R. Haynor, H. Vesselle, T.K. Lewellen, W. Eubank: PET-CT image registration in the chest using free-form deformations. IEEE Transactions on Medical Imaging, 22(1)(2003) 120-128.
[12] J.P.W. Pluim, J.B.A. Maintz, M.A. Viergever: Mutual-information-based reg-istration of medical images: a survey. IEEE Transactions on Medical Imaging, 22(8)(2003) 986-1004.
[13] D. Loeckx, P. Slagmolen, F. Maes, D. Vandermeulen, P. Suetens: Nonrigid image registration using conditional mutual information. IEEE Transactions on Medical Imaging, 29(1)(2010) 19-29.
[14] C.H. Dong, Y.-W. Chen, L.F. Lin, H.J. Hu, C.W. Jin, H.J. Yu, T. Tateyama, X.H. Han: Simultaneous Segmentation of Multiple Organs using Random Walks.
Journal of Information Processing Society of Japan, 24(2)(2015), (In Press).
[15] C.H. Dong, Y.W. Chen, T. Seki, R. Inoguchi, C.L. Lin, X.H. Han: Non-rigid Image Registration with Anatomical Structure Constraint for Assessing Locoregional Therapy of Hepatocellular Carcinoma. Journal of Computerized Medical Imaging and Graphics, 45(2015) 75-83.
[16] Y.W. Chen, K. Tsubokawa, R. Xu, S. Morikawa, Y. Kurumi: Semiautomat-ic non-rigid 3-D image registration for MR-guided liver cancer surgery. IEEE International Conference on Image Processing (ICIP), (2008) 1800-1803.
[17] Y.S. Luo, A.C.S. Chung: Nonrigid image registration with crystal dislocation energy. IEEE Transactions on Image Processing, 22(1)(2013) 229-243.
[18] L. Laura-de-Manuel, G. Wollny, J. Kybic, D. Jimenez-Carretero, J.M. Tellado, E. Ramon, M. Desco, A. Santos, J. Pascau, M.J. Ledesma-Carbayo: Organ-focused mutual information for nonrigid multimodal registration of liver CT and Gd-EOB-DTPA-enhanced MRI. Journal of Medical Image Analysis, 18(1)(2014) 22-35.
[19] P. Thevenaz, M. Unser: Optimization of mutual information for multireso-lution image registration. IEEE Transactions on Image Processing, 9(12)(2000) 2083-2099
[20] T.M. Lehmann, C. Gonner, K. Spitzer: Survey: Interpolation methods in medical image processing. IEEE Transactions on Medical Imaging, 18(11)(1999) 1049-1075.
[21] J. Kybic, M. Unser: Fast parametric elastic image registration. IEEE Trans-actions on Image Processing, 12(11)(2003) 1427-1442.
[22] J. Nocedal, S.J. Wright: Numerical optimization. J. of Springer series in op-erations research(New York).
Chapter 6 Conclusion
Medical image segmentation have been important research topics in medical im-age analysis. The work of this dissertation has mainly focused on medical imim-age segmentation and its application. In this dissertation, we proposed two segmenta-tion algorithms: one is an interactive random walks based segmentasegmenta-tion method, which can achieve a high-accuracy results but requires users to manually define seed points; the other is an automatic iterative probabilistic atlas based segmentation method, which handles more intelligently but has a drop in impact on the speed and accuracy. Finally, as an application, we developed a system to assess locore-gional therapy (LT) of hepatocellular carcinoma (HCC) by applying our proposed segmentation and registration methods. The contributions of this dissertation are summarized as following.
In the Chapter 3, we proposed an efficient interactive method based on random walks for multi-organ segmentation. A random walks (RW) method can be applied to segment the 3D medical image, however, it faced the following two problems:
One is a large computation burden due to partition of large-scale graph, Another is inaccurate segmentation because of unavailability of inappropriate initial seed points setting. In order to overcome the above problems, a knowledge-based frame-work in a slice-by-slice manner is used for organ segmentation based on RW algo-rithm. In order to automatically and efficiently segment the liver slice-by-slice, a prior knowledge of the previous segmented organ was integrated into our strategy, which included: (1) Automatic selection of seed points according to the previous segmented liver (shape constraints); (2) Refinement of seed points to the previous segmented liver (intensity constraints). Finally, a combinational RW algorithm is applied to automatically segment the whole volume in slice-by-slice manner. This slice-by-slice strategy can reduce the graph scale and significantly speed up the optimization procedure for the graph. The segmentation results demonstrated a high precision of the proposed approach. Additionally, we extend this proposed method to segment multiple organs simultaneously. Compared with conventional
RW segmentation methods, our proposed method show an improvement in the accuracy and computation time for multi-organ segmentation.
In the Chapter 4, we proposed an automatic algorithm based on iterative prob-abilistic atlas for multi-organ segmentation. Probprob-abilistic atlas based on human anatomical structure, used as a priori information in a Bayes framework, has been widely used for organ segmentation. Accurate registering of the probabilistic atlas onto the input CT volumes is difficult because of the variability in organ shape.
Moreover, the conventional probabilistic atlas may cause bias toward the specific patient study because of the single reference. As an improvement over the previous registration scheme for the probabilistic atlas, the proposed method use a template matching strategy to find an optimal geometric location of the candidate organ based on a novel iterative probabilistic atlas. First of all, a bounding box based on human anatomical localization is detected for the candidate organ and then the iterative probabilistic atlas is treated as a template to find the organ in this bounding box by the use of template matching technology. Compared with the conventional atlas-based methods, our automated method significantly improve both segmentation accuracy and robustness for multi-organ segmentation.
In the Chapter 5, we developed a system to assess locoregional therapy (LT) of hepatocellular carcinoma (HCC) by applying our proposed segmentation and registration techniques. The proposed knowledge-based of the random walks seg-mentation method can be applied to segment the livers, tumors and treated region.
Meanwhile, the proposed anatomical structure constraints based the non-rigid reg-istration method can be used to automatically register the treated region after LT to the tumor before LT. This proposed registration method extends a classical intensity-based non-rigid registration that uses an anatomical structure term to constrain the local deformation. This application enables the surgeon to have di-rectly perceived sense to evaluate the treated region of LT for HCC. The utility of our system for real clinical use was evaluated by doctors.
Publication List
Peer-reviewed Journal Papers
1. Chunhua Dong, Yen-wei Chen, Lanfen Lin, Hongjie Hu, Chongwu Jin, Hua-jun Yu, Xian-hua Han, Tomoko Tateyama. Simultaneous Segmentation of Multiple Organs using Random Walks, Journal of Information Processing Society of Japan, vol.24, no.2, 2016, (In Press).
2. Chunhua Dong, Yen-wei Chen, Amir Hossein Foruzan, Lanfen Lin, Xian-hua Han, Tomoko Tateyama, Xing Wu, Gang Xu, Huiyan Jiang. Segmentation of Liver and Spleen based on Computational Anatomy Models, Computers in Biology and Medicine, vol.67, pp.146-160, 2015.
3. Chunhua Dong, Yen-wei Chen, Toshihito Seki , Ryosuke Inoguchi, Chen-Lun Lin, Xian-hua Han. Non-rigid Image Registration with Anatomical Structure Constraint for Assessing Locoregional Therapy of Hepatocellular Carcinoma, Computerized Medical Imaging and Graphics, vol.45, pp.75-83, 2015.
4. Titinunt Kitrungrotsakul, Chunhua Dong, Tomoko Tateyama, Xian-hua Han, Yen-wei Chen. Interactive Segmentation and Visualization System for Medi-cal Images on Mobile Devices,Journal of Japan Society for Simulation Tech-nology, vol.2, no.1, pp.96-107, 2015.
5. Yen-Wei Chen, Jie Luo, Chunhua Dong, Xianhua Han, Tomoko Tateyama, Akira Furukawa, Shuzo Kanasaki. Computer-Aided Diagnosis and Quan-tification of Cirrhotic Livers based on Morphological Analysis and Machine Learning,Journal of Computational and Mathematical Methods in Medicine, vol.2013, Article ID 264809, 8 pages, 2013. doi:10.1155/2013/264809.
Peer-reviewed Conference Papers
1. Chunhua Dong, Yen-wei Chen, Amir Hossein Foruzan, Xian-Hua Han, Tomoko Tateyama, Xing Wu. A Framework for Probabilistic Atlas-based Organ
Segmentation,Proc. of Society of Photo-Optical Instrumentation Engineers (SPIE) Conference on Medical Imaging (SPIE 2016), 2016, (Accepted).
2. Chunhua Dong, Yen-wei Chen, Lanfen Lin, Hongjie Hu, Chongwu Jin, Hua-jun Yu, Tomoko Tateyama, Xian-hua Han. A Knowledge-based Interactive Liver Segmentation using Random Walks, Proc. of 12th International Con-ference on Fuzzy Systems and Knowledge Discovery (FSKD’15), pp.1765-1770, 2015.
3. Chunhua Dong, Amir Hossein Foruzan, Xian-hua Han, Tomoko Tateyama, Yen-wei Chen. Automatic Segmentation of Spleen based on Anatomical Model and Template Matching, Proc. of International Conference on Com-puter Information Systems and Industrial Applications (CISIA 2015), pp.
585-588, 2015.
4. Chunhua Dong, Toshihito Seki, Ryosuke Inoguchi, Xian-hua Han, Yen-wei Chen. CAD System for Evaluating Locoregional Therapy of Hepatocellular Carcinoma, Proc. of 29th International Congress and Exhibition on Com-puter Assisted Radiology and Surgery (CARS 2014), pp.S300-S301, 2014.
5. Yen-wei Chen, Amir Hossein Foruzan, Chunhua Dong, Tomoko Tateyama, Xian-hua Han. Automatic Segmentation of Liver from CT Images Using Probabilistic Atlas and Template Matching, Proc. of Smart Digital Futures 2014, Frontiers in Artificial Intelligence and Applications, pp.412-420, 2014.
6. Chunhua Dong, Toshihito Seki, Ryosuke Inoguchi, Chen-Lun Lin, Xian-hua Han, Yen-wei Chen. Nonrigid Registration for Evaluating Locoregional Ther-apy of Hepatocellular Carcinoma,Proc. of 6th International Conference on Biomedical Engineering and Informatics (BMEI 2013), pp.811-816, 2013.
7. Yen-wei Chen, Chunhua Dong, Xian-hua Han, Tomoko Tateyama, Shuzo Kanasaki, Akira Furukawa. Quantifying Stage Progress of Cirrhotic Livers based on Statistic Shape Models, Proc. of 6th International Conference on Biomedical Engineering and Informatics (BMEI 2013), pp.822-825, 2013.
Conference Papers without review
1. Yingbo Li, Chunhua Dong, Tomoko Tateyama, Yen-wei Chen. Liver Segmen-tation Using Iterative Probabilistic Atlas and Template Matching Technique, IEICE Technical Report, MI2015-34, vol.115, no.139, pp.13-17, 2015.
2. Chunhua Dong, Amir Hossein Foruzan, Xian-hua Han, Tomoko Tateyama, Yen-wei Chen. Organ Bounding Box Annotation based on Adaptive Se-lection of Bone References, IEICE Technical Report, MI2014-25, vol.114, no.103, pp.311-315, 2014.
3. Titinunt Kitrungrotsakul, Chunhua Dong, Xian-hua Han, Yen-wei Chen.
Improved Interactive Medical Image Segmentation Using Graph Cut and Superpixels,IEICE Technical Report, MI2014-25, vol.114, no.103, pp.17-20, 2014.