MRI Liver Tumor Classification Using Machine Learning Approach and Structure Analysis

 

Shyamala Devi M*, Sruthi A. N, Saranya Jothi C

Associate Professor, Assistant Professor, Department of Department of Computer Science and Engineering,

Vel Tech Rangarajan Dr.Saguntahala R and D Institute of Science and Technology, Avadi, Chennai, TN, India.

*Corresponding Author E-mail: shyamalapmr@gmail.com

 

ABSTRACT:

The survival rate of Liver tumor patients can be improved if we perform early detection and treating them. The clinical researches have exposed that the volume measurement can give the best reflection of the tumor response. The liver tumor requires the tumor segmentation. This paper proposes an automatic support system for stage classification using artificial neural network (learning machine) and to detect Liver Tumor through fuzzy clustering methods for medical application. The detection of the Liver Tumor is a challenging problem, due to the structure of the Tumor cells. This project presents a segmentation method, fuzzy clustering algorithm, for segmenting Magnetic Resonance images to detect the Liver Tumor in its early stages and to analyze anatomical structures.

 

KEYWORDS: Liver Tumor, ANN, Segmentation, Fuzzy Clustering, Resonance Image.

 


1. INTRODUCTION:

Automated classification and detection of Tumor in different medical images is motivated by the necessity of high accuracy when dealing with a human life. It has been proven that double reading of medical images could lead to better Tumor detection. But the cost implied in double reading is very high, that’s why good software to assist humans in medical institutions is of great interest nowadays.  Conventional methods of monitoring and diagnosing the diseases rely on detecting the presence of particular features by a human observer.  Several techniques for automated diagnostic systems have been developed in recent years to attempt to solve this problem. Such techniques work by transforming the mostly qualitative diagnostic criteria into a more objective quantitative feature classification problem. In this paper, the automated classification of Liver magnetic resonance images by using some prior knowledge like pixel intensity and some anatomical features is proposed.

 

The high frequency incidence and mortality values reveals how deadly liver tumor is progressing. Large tumors can be simply recognized by imaging techniques. However, late recognition does not save life. In order to recover survival, liver tumor has to be detected in an early stage.

 

2. RELATED WORK:

In Liver disease diagnosis system, presence of less accuracy in an automatic image classification and segmentation of tissue analysis due to poor texture discrimination and clustering inefficiency. To overcome this limitation, here multi scale decomposition based texture analysis and spatial clustering will be used for better classification and segmenting tissues. Fuzzy c-means (FCM) algorithm is one of the most widely used fuzzy clustering algorithms in image segmentation. FCM algorithm was first introduced by Dunn [6] and later extended by Bezdek [7]. Although the conventional FCM algorithm works well on most noise-free images, it fails to segment images corrupted by noise, outliers and other imaging artifacts. A lot of clustering based methods has been proposed for image segmentation. Among the clustering methods, one of the most popular methods for image segmentation is fuzzy clustering, which can retain more image information than hard clustering Ahmed et al. [8] proposed FCM_S, which modified the objective function of FCM by introducing the spatial neighborhood term. One drawback of FCM_S is that the spatial neighborhood term is computed in each iteration step, which is very time-consuming. To speed up the image segmentation process, Szilagyi et al. [10] proposed the enhanced FCM (EnFCM), which form a linearly-weighted sum image from both the local neighborhood average gray level of each pixel and original image, and then clustering is performed on the basis of the gray level histogram of summed image. Stelios et al. [1] presents a novel robust fuzzy local information c-means clustering algorithm (FLICM), which is free of any parameter selection, as well as promoting the image segmentation performance. In FLICM, a novel fuzzy factor is defined to replace the parameter a used in above algorithms and its variants. More recently, we [12] proposed a variant of FLICM algorithm (RFLICM), which adopts the local coefficient of variation to replace the spatial distance as a local similarity measure. Furthermore, it presents a more robust result. Particularly, the clustering algorithms based on the kernel methods have been shown to be robust to the outliers or noises of the dataset [17].

 

2.1. Existing Techniques

The Existing Techniques are as follows,

·        Thresholding method

·        K means clustering 

·        Manual analysis - time consuming, inaccurate and requires intensive trained person to avoid diagnostic errors.

 

2.2. Drawback of the Existing Techniques

The drawback of the Existing Techniques are as follows,

·        Difficult to get accurate results

·        Not applicable for multiple images for Tumor detection in a short time

·        Medical Resonance images contain a noise caused by operator performance which can lead to serious inaccuracies classification.

 

3. PROPOSED WORK:

3.1. Multilevel Wavelet Decomposition:

A wavelet series is a representation of a square integrable (real-or complex-valued) function by a  certain orthonormal series generated by a wavelet. Here we are using four level wavelet decomposition. This article provides a formal, mathematical definition of an orthonormal wavelet and of the integral wavelet transform. The wavelet transform has gained widespread acceptance in signal processing in general and in image compression research in particular.

 

3.2. Contributions:

In applications such as still image compression, discrete wavelets transform (DWT) based schemes have outperformed other coding schemes like the ones based on DCT. Since there is no need to divide the input image into non-overlapping 2-D blocks and its basis functions have variable length, wavelet-coding schemes at higher compression ratios avoid blocking artifacts. Because of their inherent multi -resolution nature, wavelet-coding schemes are especially suitable for applications where scalability and tolerable degradation are important. Recently the JPEG committee has released its new image coding standard, JPEG-2000, which has been based upon DWT. The classification of MRI Liver image is shown below in fig 1.

 


 

Fig.1: Classification of MRI Liver Image

 


3.3. Feature Extraction

Energy:

It is a gray-scale image texture measure of homogeneity changing, reflecting the distribution of image gray-scale uniformity of weight and texture.

E =

 

 

p(x,y) is the GLC M

 

Contrast:

Contrast is the main diagonal near the moment of inertia, which measure the value of the matrix is distributed and images of local changes in number, reflecting the image clarity and texture of shadow depth.

 

Contrast I =

 

 

Entropy:

It measures image texture randomness, when the space co-occurrence matrixes for all values are equal, it achieved the minimum value.

 

S =

 

 

Homogeneity:

Measures the closeness of the distribution of elements in the GLCM to the GLCM diagonal

Homogeneity = sum (sum (p(x, y) / (1 + [x-y])))

 

3.4. Supervised Classifier:

Supervised learning with non knowledge based classifier will be used for gender classification. he neural network model PNN is used here to act as a classifier with radial basis function for network activation function. The training samples features with assigned target vectors are fed into created PNN model for supervised training to get network parameters such as node biases and weighting factors. Finally, test image features are simulating with trained network to make decision of gender Male or Female. The general architecture of neural network is shown in fig 2.

 

 

Fig. 2: General architecture of Neural Network

 

Probabilistic networks perform classification where the target variable is categorical and this architecture has three types of layers such as an input layer, a pattern layer, and an output layer. Here is one neuron in the input layer for each predictor variable. In the case of categorical variables, N-1 neurons are used where N is the number of categories. The input neurons(or processing before the input layer) standardizes the range of the values by subtracting the median and dividing by the interquartile range. The input neurons then feed the values to each of the neurons in the hidden layer.

 

There is one pattern neuron for each category of the target variable. The actual target category of each training case is stored with each hidden neuron; the weighted value coming out of a hidden neuron is fed only to the pattern neuron that corresponds to the hidden neuron’s category. The pattern neurons add the values for the class they represent (hence, it is a weighted vote for that category).

 

The decision layer is different for PNN and GRNN networks. For PNN networks, the decision layer compares the weighted votes for each target category accumulated in the pattern layer and uses the largest vote to predict the target category.

 

3.5. Segmentation:

Spatial Fuzzy clustering:

Clustering is used to classify items into identical groups in the process of data mining. It also exploits segmentation which is used for quick bird view for any kind of problem. K-Means is a well known partitioning method. Objects are classified as belonging to one of k groups, k chosen a priori. Cluster membership is determined by calculating the centered for each group and assigning each object to the group with the closest centroid. This approach minimizes the overall within-cluster dispersion by iterative reallocation of cluster members.

 

Based on this, K-means segmentation of Alzheimer’s Disease in MRI scan datasets is implemented. This produces the proven results for PET scan datasets using K-Means clustering. But in same datasets, if different structures exist, ithas often found to fail. In general, the fuzzy c-means algorithm is assigned the pixels to fuzzy clusters without any label. Hard clustering methods are used to group pixels to belong exclusively one cluster. But, FCM allows a pixel in more than one cluster depends on the degrees of membership.

 

Summation of membership of each data points in the given datasets should be equal to each other. Let X ={x1, x2, x3 ..., xn} be the set of data points and C = {c1, c2, c3 ..., cn} be the set of centers. The following equations 1and 2 explain the membership and cluster center updation for each iteration. One of the important characteristics of an image is that neighboring pixels are highly correlated. In other words, these neighboring pixels possess similar feature values, and the probability that they belong to the same cluster is great. This spatial relationship is important in clustering, but it is not utilized in a standard FCM algorithm. The probability that they belong to the same cluster is great. This spatial relationship is important in clustering, but it is not utilized in a standard FCM algorithm. The SFCM algorithm is shown below.

 

Algorithm 1: SFCM( )

--------------------------------------------------------------------

Input: MRI Liver image Read data from MRI scan image

1.      Randomly select the number of clusters

2.      Initialize the fuzziness factor

3.      For i=1:max_iter

4.      Calculate the distance between pixel and centroid

5.      Calculate the membership values

6.      End

7.      If distance < threshold

8.      Update the membership values

9.      Update cluster centroid

10.   Else

11.   Determine objective function

12.   End if

13.   Do the Segmentation

--------------------------------------------------------------------

 

4. RESULTS AND DISCUSSION:

Probabilistic Neural Network with radial basis function will be employed to implement an automated Liver tumor classification. Decision making was performed in two stages: feature extraction using GLCM and the classification using PNN-RBF network. The performance of this classifier was evaluated in terms of training performance and classification accuracies. The simulated results will be shown that classifier and segmentation algorithm provides better accuracy than previous method. The segmented tumor area will be determined by counting the number of non-zero pixels and one pixel’s area (0.264 mm).

Area = sqrt (Pixel Count) * 0.264 mm. ^2

 

Fig. 3: Input MRI Images

 

Fig. 4: Multilevel Wavelet Decomposition

 

Fig. 5: Segmented Image

 

Fig. 6: Area of Affected Region

 

Fig. 7: Result Validation of Liver Tumor

 

5. CONCLUSION:

This paper deals with automated diagnosis classification system from processing of MRI Liver image for bio medical application. This processing system could provide the results either normal or abnormal with better classification accuracy. It worked effectively with an influence of methodologies such as multilevel wavelet decomposition, GLCM for texture analysis and supervised probabilistic neural network which acts as a classifier. Here spatial fuzzy clustering algorithm was utilized effectively for accurate tumor detection to measure the area of abnormal region. From an experiment, system proved that it provides better classification accuracy with various stages of test samples and it consumed less time for process. Our future work is to examine the proposed methodology to check the validity of other MRI disease images.

 

CONFLICT OF INTEREST:

The authors declare no conflict of interest.

 

REFERENCES:

1.       N. Kwak, and C. H. Choi, “Input Feature Selection for Classification Problems,” IEEE Transactions on Neural Networks, 13(1), 143–159, 2002.

2.       E. D. Ubeyli and I. Guler, “Feature Extraction from Doppler Ultrasound Signals for Automated Diagnostic Systems”, Computers in Biology and Medicine, 35(9), 735–764, 2005.

3.       D.F. Specht, “Probabilistic Neural Networks for Classification, mapping, or associative memory”, Proceedings of IEEE International Conference on Neural Networks, Vol.1, IEEE Press, New York, pp. 525-532, June 1988.

4.       D.F. Specht, “Probabilistic Neural Networks” Neural Networks, vol. 3, No.1, pp. 109-118, 1990.

5.       Georgiadis. Et al , “ Improving Liver tumor characterization on MRI by probabilistic neural networks and non-linear transformation of textural features”, Computer Methods and program in biomedicine, vol 89, pp24-32, 2008

6.       Kaus M., “ Automated segmentation of MRI Liver tumors”, Journal of Radiology, vol. 218, pp. 585-591,2001

7.       J. Dunn, “A fuzzy relative of the ISODATA process and its use in detecting compact well-separated clusters,” J. Cybern., vol. 3, no. 3,pp. 32–57, 1974.

8.       J. Bezdek, Pattern Recognition with Fuzzy Objective Function Algorithms. New York: Plenum, 1981.

9.       M. Ahmed, S. Yamany, N. Mohamed, A. Farag, and T. Moriarty, “A modified fuzzy C-means algorithm for bias field estimation and segmentation of MRI data,” IEEE Trans. Med. Imag., vol. 21, no. 3,pp. 193–199, Mar. 2002.584 IEEE TRANSACTIONS ON IMAGE PROCESSING, VOL. 22, NO. 2, FEBRUARY 2013

10.     S. Chen and D. Zhang, “Robust image segmentation using FCM with spatial constraints based on new kernel-induced distance measure,” IEEE Trans. Syst., Man, Cybern., B, Cybern., vol. 34, no. 4, pp. 1907–1916,Aug. 2004.

11.     L. Szilagyi, Z. Benyo, S. Szilagyii, and H. Adam, “MR brain image segmentation using an enhanced fuzzy C-means algorithm,” in Proc.25th Annu. Int. Conf. IEEE EMBS, Nov. 2003, pp. 17–21.

12.     W. Cai, S. Chen, and D. Zhang, “Fast and robust fuzzy C-means clustering algorithms incorporating local information for image segmentation,” Pattern Recognit., vol. 40, no. 3, pp. 825–838, Mar.2007.

13.     M. Gong, Z. Zhou, and J. Ma, “Change detection in synthetic aperture radar images based on image fusion and fuzzy clustering,” IEEE Trans. Image Process., vol. 21, no. 4, pp. 2141–2151, Apr.2012.

14.     N. Cristianini and J. S. Taylor, An Introduction to SVM’s and Other Kernel-Based Learning Methods. Cambridge, U.K.: Cambridge Univ. Press, 2000.

15.     X. Yang and G. Zhang, “A kernel fuzzy C-means clustering-based fuzzy support vector machine algorithm for classification problems with outliers or noises,” IEEE Trans. Fuzzy Syst., vol. 19, no. 1, pp. 105–115,Feb. 2011.

16.     V. Roth and V. Steinhage, “Nonlinear discriminant analysis using kernel functions,” in Advances in Neural Information Processing Systems 12, S. A Solla, T. K. Leen, and K.-R. Muller, Eds. Cambridge, MA: MIT Press, 2000, pp. 568–574.

17.     B. Scholkopf, A. J. Smola, and K. R. Muller, “Nonlinear component analysis as a kernel eigenvalue problem,” Neural Comput., vol. 10, no. 5, pp. 1299–1319, 1998.

18.     K. Wu and M. Yang, “Alternative C-means clustering algorithms,” Pattern Recognition, vol. 35, no. 10, pp. 2267–2278, 2002.

 

 

 

 

 

Received on 03.12.2017          Modified on 28.12.2017

Accepted on 15.01.2018        © RJPT All right reserved

Research J. Pharm. and Tech 2018; 11(2):434-438.

DOI: 10.5958/0974-360X.2018.00080.X