An Efficient Hybrid Segmentation Algorithm for Computer Tomography Image Segmentation
IV. Experimental Results and Discussion
Experimentation was carried out on 100 numbers of different tumor patients contains 100 to 1000 slices of Computer Tomography images using Segmentation algorithms. The image format is DICOM (Digital Imaging Communications in Medicine). The algorithm has been implemented in Matlab environment. Manual Segmentation done by the medical expert. Experimental results of the images are illustrated here. Fig. 1(a) depicts input CT image. Fig. 1(b) describes segmentation result generated with HMSK. Fig. 1(c) depicts SQFD algorithm on the CT images. Fig. 1(d) shows Hybrid Segmentation result. Fig. 2(a) shows segmentation result generated with HMSK for Liver region. Fig. 2(b) depicts results generated with SQFD algorithm for Liver Region. Fig. 2(c) shows segmentation result generated with proposed Hybrid segmentation. Fig. 2(d) depicts the manual segmentation results contoured by the medical experts for liver organ.
IV.1. Performance Analysis of HMSK, SQFD, Hybrid Segmentation Method
Selecting the suitable segmentation evaluation measure is a complex task. A variety of performance
measures to assess the medical image segmentation methods are available in present scenario. Generally sensitivity, specificity and accuracy are used to evaluate the segmentation methods in a good manner.
Fig. 1(a). Input Image Fig. 1(b). HMSK Segmentation
Fig. 1(c). SQFD Segmentation Fig. 1(d). Hybrid Segmentation
Liver Segmentation Output
Fig. 2(a). HMSK Segmentation Fig. 2(b). SQFD Segmentation output
Fig. 2(c). Hybrid Segmentation output
Fig. 2(d). Manual Segmentation done by the experts They are defined as:
Sensitivity = TP TP TN Specificity= TN TNFP Accuracy= TN TP TP TN FP FN
V. V. Gomathi, S. Karthikeyan
where TP (True Positive) is the number of pixels of the foreground that are correctly classified, TN (True Negative) is the number of pixels of the background that are correctly classified, FP (False Positive) is the number of pixels of the background that are classified as foreground and FN (False Negative) is the number of pixels of the foreground that are Classified as background.
Accuracy refers to the degree to which the segmentation results agree with the true segmentation i.e. Correct segmented pixels in the object. Fragments indicate that the number of connected components in the required region to identify as organ.
In this paper we also consider the fragments parameter. If more number of fragments exists in the image, the segmentation task is also complicated.
TABLEI
PERFORMANCE ANALYSIS OF HMSK,SQFD, HYBRID SEGMENTATION ALGORITHM
Quantitative Parameters CT Image Segmentation Algorithm HMSK SQFD Hybrid Segmentation (HMSK with SQFD) Sensitivity 93.19 98.12 98.12 Specificity 99.26 99.99 99.95 Accuracy 99.05 99.22 99.65 Number of fragments 350 173 133
We have proposed and successfully implemented a new integrated method for segmenting real time Computer tomography images. This paper mainly concentrates to propose a new algorithm with comparison of HMSK and SQFD algorithm.
In our previous research, we have compared many segmentation algorithms. The segmentation results of HMSK and SQFD are considered the best algorithm that gives better segmentation results for real CT images. The HMSK algorithm is based on clustering based segmentation algorithm and SQFD algorithm is a distance based algorithm.
In cluster based medical image segmentation algorithms, more number of unwanted fragments present and also fragments are not consistent when executed for a certain number of times i.e. when the same image executed for different number of times, the result were not holding the same number of fragments, position of fragment and size of fragment and also were dynamic.
For diminishing these drawbacks, the distance based segmentation algorithm has been proposed. In our previous distance based research, we have compared five distance measures namely Euclidean distance, Manhattan Distance, Minkowski distance, Chebyshev distance and Signature Quadratic form Distance(SQFD) measures[21]. The SQFD algorithm finds the similarity between all cluster elements and every pixel values such that the feature space have the highest possible similarity values of cluster vector. One of the drawback exist in this SQFD is Still number of fragments exist is more. The HMSK and SQFD algorithm is also not good fit for exact computer tomography image segmentation. Hence we proposed a Hybrid Segmentation method.
88 90 92 94 96 98 100
Sensitivity Specificity Accuracy
Performance Analysis of HMSK, SQFD, Hybrid Segmentation Algorithm
HMSK SQFD
Hybrid Segmentation
Fig. 3(a). Performance analysis of HMSK, SQFD, Hybrid segmentation algorithms in terms of sensitivity, specificity and accuracy
0 100 200 300 400 Number of fragments
Perform ance Analysis of HMSK,SQFD, Hybrid segm entation Algorithm
HMSK
SQFD Hybrid Segmentation
Fig. 3(b). Performance analysis of HMSK, SQFD, Hybrid segmentation algorithms in terms of number of fragments
There are more number of quantitative metrics are available to evaluate the segmentation accuracy. Here four important parameters are used to determine the precision of the HMSK, SQFD and Hybrid segmentation algorithms such as Sensitivity, specificity, accuracy and number of fragments. The important parameters mentioned above i.e. Sensitivity, specificity, accuracy and number of fragments are specified in Table I, for the HMSK, SQFD, Hybrid segmentation algorithms. In this paper, the liver organ has taken for demonstrating the segmentation performance. Segmentation of liver is a tedious process.
Liver is joined with some other organs such as heart, lung, stomach, spleen etc. In Figs. 2(a), (b) and (c) describes the liver segmentation done by HMSK, SQFD and Hybrid Segmentation algorithm.
Hybrid Segmentation algorithm having less number of fragments compared to HMSK and SQFD algorithm based on the visible part and analysis part. The segmentation result is compared with manual contoured image already done by the medical expert. Manual segmentation result is represented in Fig. 2(d).
The proposed approach is precise, robust and provides good quality results. From the experimental results obtained it can be concluded that our proposed Hybrid segmentation algorithm performs well in segmenting the real CT images with less number of fragments (connected components). The liver and heart organs are well separated and heart and spleen are also well
V. V. Gomathi, S. Karthikeyan
Copyright © 2014 Praise Worthy Prize S.r.l. - All rights reserved International Review on Computers and Software, Vol. 9, N. 9 separated. The proposed hybrid segmentation method
gives the highest accuracy than other methods.
V.
Conclusion
In this paper, a hybrid segmentation algorithm is proposed to get the precise fragments from Computer tomography images. We found that the clustering with distance based algorithm is well suitable for exact computer tomography image segmentation.
The method integrates medoidshift with K-means algorithm and Signature quadratic form distance algorithm. Such integration reduces the drawbacks of both the methods. The benefit of the methodology is that it produces high quality segmentations of Computer tomography images and also separates the joined organ efficiently.
The computational results show that the proposed methodology achieves high quality precise segmentation results with less number of fragments.
Hence the complexity is reduced for exact organ classification.
References
[1] Dinesh D. Patil, Sonal G. Deore, Medical Image Segmentation: A Review, International Journal of Computer Science and Mobile
Computing, Vol. 2, n.1, pp.22 – 27, 2013.
[2] Neeraj Sharma and Lalit M. Aggarwal, Automated medical image segmentation technique. Journal of Medical Physics, Vol.35, n.1, pp.3–14, 2010.
[3] Ladak HM, Mao F, Wang Y, Downey DB, Steinman DA, Fenster A., Prostate boundary segmentation from 2D ultrasound images,
Journal of Medical Physics, Vol.27,n.8,pp.1777-1788,2000.
[4] Yiqiang Zhan and Dinggang Shen, Deformable Segmentation of 3-D Ultrasound Prostate Images Using Statistical Texture Matching Method, IEEE Transactions on Medical Imaging, Vol.
25, n. 3, pp.256-272, 2006.
[5] Djamal Boukerroui , Atilla Baskurt c, J. Alison Noble , Olivier Basset , Segmentation of ultrasound images––multiresolution 2D and 3D algorithm based on global and local statistics, Pattern Recognition Letters,Vol.24 , pp.779–790, 2003.
[6] Ashish Thakur Radhey Shyam Anand, A Local Statistics Based Region Growing Segmentation Method for Ultrasound Medical Images, International Journal of Medical, Health, Pharmaceutical and Biomedical Engineering Vol.1, n.10, pp.570-
575, 2007.
[7] Elnomery Zanaty and Sultan Aljahdali, Improving Fuzzy Algorithms for Automatic Magnetic Resonance Image Segmentation, The International Arab Journal of Information
Technology, Vol. 7, n.3, pp.271-279, 2010.
[8] Mohamed N. Ahmed, Sameh M. Yamany, Nevin Mohamed, Aly A. Farag, and Thomas Moriarty, A Modified Fuzzy C-Means Algorithm for Bias Field Estimation and Segmentation of MRI Data, IEEE Transactions on Medical Imaging, Vol. 21, n. 3, pp.193-199, 2002.
[9] Jianzhong Wangm, Jun Kong, Yinghua Lu,,Miao Qi, Baoxue Zhanga, A modified FCM algorithm for MRI brain image segmentation using both local and non-local spatial constraints,
Computerized Medical Imaging and Graphics, Vol. 32, n.8,
pp.685–698, 2008.
[10] Paresh Chandra Barman, Sipon Miah, Bikash Chandra Singh and Mst. Titasa Khatun, MRI mage segmentation using level set method and implement an medical Diagnosis system, Computer
Science & Engineering: An International Journal (CSEIJ), Vol.1,
n.5, pp.1-10, 2011.
[11] Shan Shen, William Sandham, Member, IEEE, Malcolm Granat,
and Annette Sterr , MRI Fuzzy Segmentation of Brain Tissue Using Neighborhood Attraction With Neural-Network Optimization, IEEE Transactions On Information Technology In
Biomedicine, Vol. 9, n.3, pp.459-467, 2005.
[12] Iraky khalifa , Aliaa Youssif , Howida Youssry, MRI Brain Image Segmentation based on Wavelet and FCM Algorithm,
International Journal of Computer Applications, Vol.47, n.16,
pp.32-39, 2012.
[13] Chung-Yi Huang, Lai-Jun Luo, Pei-Yuan Lee, Jiing-Yih Lai,,Wen-Teng Wang, Shang-Chih Lin, Efficient Segmentation Algorithm for 3D Bone Models Construction on Medical Images,
Journal of Medical and Biological Engineering, Vol.31, n.6,
pp.375-386, 2010.
[14] A .Morenoa, C.M. Takemuraa, O .Colliotc ,O .Camarad, I .Blocha, Using anatomical knowledge expressed as fuzzy constraints to segment the heart in CT images, Pattern
Recognition,Vol.41,n.8, pp. 2525 – 2540, 2008.
[15] X.M. Pardo. , M.J. Carreira , A. Mosquera, D. Cabello, A snake for CT image segmentation integrating region and edge information, Image and Vision Computing,Vol.19, n.7, pp.461- 475, 2001.
[16] Zikuan Chen, Sabee Molloi, Automatic 3D vascular tree construction in CT angiography, Computerized Medical Imaging
and Graphics,Vol.27, pp.469–479,2003.
[17] Yaser Ajmal Sheikh, Erum Arif Khan, Takeo Kanade, Mode- seeking by Medoidshifts. Computer Vision (ICCV), IEEE International conference on, pp.1-8,2007.
[18] V.V.Gomathi, Dr.S.Karthikeyan, An Efficient Clustering based Segmentation Algorithm for Computer Tomography Image Segmentation, Journal of biomedical engineering and medical
imaging, vol.1, n.3, pp. 1-11, 2014.
[19] J.L Marroquin, F. Girosi, Some Extentions of the K-Means Algorithm For Image Segmentation and Pattern Classification, Technical Report, MIT Artificial Intelligence Laborartory,1993. [20] M.Luo, Y.F.Ma ,H.J. Zhang, A Special Constrained K-Means
approach to Image Segmentation ,proceedings of the Fourth
International Conference on Information Communications and Signal Processing and the Fourth Pacific Rim Conference on Multimedia,Vol.2,pp.738-742,2003.
[21] V.V. Gomathi, S. Karthikeyan, Performance Analysis of Distance Measures for Computer tomography Image Segmentation, International Journal of Computer Technology
and Applications, Vol. 5, n.2, pp. 400-405,2014.
[22] Beecks.C, Uysal M.S, Seidl.T, Signature Quadratic Form Distances for Content-based Similarity, ACM CVIR 2010. [23] V.V. Gomathi , S. Karthikeyan, A Proposed Hybrid Medoid Shift
with K-Means (HMSK) Segmentation Algorithm to Detect Tumor and Organs for Effective Radiotherapy, Lecture Notes in
Computer Science(Springer), Vol. 8284, pp.139-147, 2013.
[24] Ebrahim, M.J., Pourghassem, H., A novel automatic synthetic segmentation algorithm based on mean shift clustering and canny edge detector for aerial and satellite images, (2012) International
Review on Computers and Software (IRECOS), 7 (3), pp. 1122-
1129.
[25] Ali Hassan Al-Fayadh, Hind Rostom Mohamed ,Raghad Saaheb Al-Shimsah, CT Angiography Image Segmentation by Mean Shift Algorithm and Contour with Connected Components Image,
International Journal of Scientific & Engineering Research, Vol.3, n. 8, pp.1-5, 2012.
[26] Keh-Shih Chuang , Hong-Long Tzeng , Sharon Chen , Jay Wu , Tzong-Jer Chen, Fuzzy c-means clustering with spatial information for image segmentation, Computerized Medical
Imaging and Graphics,Vol.30,n.1, pp. 9–15, 2006.
[27] Wenbing Tao, Hai Jin, Yimin Zhang, ―Color Image Segmentation Based on Mean Shift and Normalized Cuts, IEEE
Transactions on systems, man, and cybernetics—part b: Cybernetics, Vol. 37, n. 5, pp.1382-1389, 2007.
[28] Zhou Wang and Alan C. Bovik, Ligang Lu, Why is image Quality Assessment So Difficult.
[29] Zhou Wang, Member, Alan C. Bovik, Image Quality Assessment: From Error Visibility to Structural Similarity, IEEE Transactions
V. V. Gomathi, S. Karthikeyan
Authors’ information
1
Ph.D Research Scholar, Research and Development Centre, Bharathiar University, Coimbatore, Tamilnadu, India.
2Assistant Professor, Department of Information Technology, College of Applied Sciences, Sohar, Oman.
V. V. Gomathi Completed MCA in