You are on page 1of 6

1

JOURNAL OF COMPUTING, VOLUME 3, ISSUE 3, MARCH 2011, ISSN 2151-9617 HTTPS://SITES.GOOGLE.COM/SITE/JOURNALOFCOMPUTING/ WWW.JOURNALOFCOMPUTING.ORG

Intelligent Diagnostic System for the diagnosis and prognosis of Breast Cancer using ANN
R.R.Janghel , Anuj Mehra , Anupam Shukla , Ritu Tiwari Soft Computing and Expert System Laboratory ABV-Indian Institute of Information Technology and Management, Gwalior, India
Abstract- Breast cancer occurs mostly in women and figures states that almost one out of eight women suffers from this disease. Breast cancer is second only to lung cancer that causes the majority of deaths among women and is difficult to identify accurately and treat. Diagnostic computational tools can be used effectively, with high degree of accuracy, which may help in improving the specificity and sensitivity of diagnoses. There are two known types of breast cancer namely benign and malignant which are cured mostly by mammography, FNA (Fine Needle Aspirate) and surgical biopsy. In this research work, our main objective is to develop and compare a diagnostic system for diagnosis; prognosis and prediction of breast cancer with the usage of the computational intelligence namely Back Propagation (BPA), Learning vector Quantization (LVQ), Support vector machine (SVM) and Adaptive Neuro-fuzzy Inference System (ANFIS). These modeling tools can highlight the important features that play pivotal roles in the classification and aid physicians to diagnose and prognosticate breast cancer. The database used in developing the above discussed model in this research work is adopted from the University of Wisconsin (UCI) Machine Learning Repository. Experimental result shows that different models give optimal performance for the data set. However, all the models are able to solve the problem up to a greater extent but a maximum accuracy of 96.50% is being achieved on the usage of LVQ which outdo all the other models. Keywords- Artificial Neural Networks, Breast Cancer, Expert System, vector machine and Neuro-fuzzy Inference System.
1 2 3 4

2. Literature Review
Yuanjiao et al. proposed a technology to extract micro-calcifications clusters with accurate edge effects which can't be detected by the naked eye on mammograms to obtain much more hidden information in order to help the doctors in diagnosing breast cancer [2]. Another important research was been carried out by Heng-Da et al. who has proposed a computerized micro-calcification detection based on fuzzy logic, vibro-acoustography and probabilistic neural network on mammograms for breast cancer [3]. Image feature extraction was utilized to retrospectively analyze screening mammograms taken prior to the detection of a malignant mass was done by Mohd. Sameti et al.[4]. Al Mutazet al. uses statistical texture features for breast cancer detection and few ANN [5]. T.Z. Tan et al.uses thermogram which is a promising front-line screening tool to warn women of cancer [6]. Laufer et al. proposed a modified self-organizing map with nonlinear weight adjustments to reduce number of unnecessary biopsies with a minimal number of subsequent [7]. Taio et al. used Kohonen self organizing map and multilayer perceptron trained with the

1. Introduction
Breast cancer has become one of the most perilous carcinomas in the world among middle-aged and old age women. Breast cancer is the second leading cause of cancer deaths in women worldwide and occurs in nearly one out of eight women [1]. Breast cancer occurs mostly in women but hardly ever it is found to occur in men. Computational intelligence is a stimulating field and from the last decade, Artificial Neural Networks (ANN) has become important in medical applications due to their accuracy for predictive inference, with potential to support large data sets. Feedforward Neural Network (FNN) is a kind of ANN, which has a better structure but due to slow training rate, easy to trap local minimum point and poor ability in global search prevents its usage as alone. Learning involves the extraction of rules or patterns from the set of data. Further the time and memory requirement would be reasonably less as the system has already summarized the data into some patterns or rules. In the computer-aided decision system, a system is considered as effective and accurate if it shows a very high generalizing capability which would help a physician to diagnose a patient correctly. However, unfortunately clinicians can act in different ways depending on their knowledge and experience which highlighted the need to introduce diagnostic tools to support the scientific homogeneity and accountability of healthcare decisions and actions. The benefits expected from such actions include an overall reduction in cost, improved quality
Back Propagation Algorithm, Learning vector Quantization, Support

of care as well as patient and public opinion satisfaction. Recently, computer-based medical data processing research has yielded methods and tools for managing the task away from the hospital management level and closer to the desired disease and patient management level.

JOURNAL OF COMPUTING, VOLUME 3, ISSUE 3, MARCH 2011, ISSN 2151-9617 HTTPS://SITES.GOOGLE.COM/SITE/JOURNALOFCOMPUTING/ WWW.JOURNALOFCOMPUTING.ORG

backpropagation algorithm [8]. Xiong et al. used statistics methods like PCA, PLS linear regression analysis, data mining methods, and hybrid system of combination rough set and probabilistic neural network. Probabilistic neural network to perform supervised classification and rough sets was able to reduce the number of attributes in the dataset without sacrificing classification accuracy in [9, 10, 11]. Karabatak et al. used ANN and ANFIS with linear discriminant analysis and principal component analysis for diagnosis of breast cancer. [12, 13]. Ky Van Ha et al. constructed a hierarchical evolutionary RBF network and employed it to detect the breast cancer. [15, 16]. The optimum network for classification of breast cancer cells was found using Hybrid Multilayer Perceptron (HMLP) network. [17, 18]. Jain et al. used fuzzy-logic, Hybrid Neuro-Fuzzy generator based on the Knowledge Oriented Design (KOD) concept and Cooperative Neuro-Fuzzy systems using Genetic Algorithms were used for the classification (diagnosis) of breast cancer [19, 20]. Pena-Reyes et al. proposed fuzzy-genetic breast cancer identification. [21]. Seker et al. proposed a methodology with neural network, fuzzy logic, FK-NN and statistical method to prognostic analysis of cytometric image data. [22].

Collect Cancer Data Processing

data

set

for

Breast

Feature Extraction and analysis. Removal OR Filling of missing data. Data Normalization.

Simulated Models Benign Malignant

Performance Comparison of computational models

Figure 1: Block Diagram of Complete Methodology

3.1. Diagnosis using Back Propagation Algorithm


In back propagation, the partial derivatives of the cost function with respect to the free parameters of the network is being determined by back propagation error signals through the network, layer by layer. The feed-forward neural network architecture used in this experiment consists of a single hidden layer along with input and output. The employed neural network is a feed forward neural network and it is shown in the following figure

3. Methodology
In this research work, computational intelligence models have been used for the diagnosis and prediction of breast cancer. The learning takes place through an iterative process of weight adjustments applied to its initial weight after epoch iteration of the learning process. Figure 1, shows the overall diagram of complete methodology. Mammography, biopsy and FNA database of breast cancer microscopic and clinical tests reports are used in our diagnostic system which helps in prediction of cancer as either Benign or Malignant. In this research work, we have employed ANN like Back Propagation (BPA), Support vector machine (SVM), learning vector Quantization (LVQ) and Adaptive Nero-fuzzy Inference System (ANFIS) for training of the dataset. Among all the above mentioned ANN, LVQ achieved the 96.50% of accuracy which comes out to be highest.
Training and Testing Phase

Figure 2. Back Propagation Multi-layer neural network structure. The transfer function in hidden layer neurons and output layer neurons are sigmoid and purelin respectively. The performance function used was Mean Sum-squared Error (MSE). For hidden units, we must propagate the error back from the output nodes; equation (1) gives us the equation of error back propagation

dE dnet,
1Pj

Sj

dnett dy, dnet,


= -2/

(1)

JOURNAL OF COMPUTING, VOLUME 3, ISSUE 3, MARCH 2011, ISSN 2151-9617 HTTPS://SITES.GOOGLE.COM/SITE/JOURNALOFCOMPUTING/

3
WWW.JOURNALOFCOMPUTING.ORG

where, first factor represents the error of node i, second factor is the weight from unit j to i, and third factor is the derivative of node j's activation function [23].

3.2. Diagnosis using Learning Vector Quantization Network


LVQ network are based on supervised classification algorithm. LVQ networks are designed to make decision boundaries in the input space, for a particular set of training data. According to structure, the network contains an input layer, a competitive layer and an output layer.

function to a single-valued output or a decision associated with the output. ANFIS are a class of adaptive networks that are functionally equivalent to fuzzy inference systems. These Layer 1 represent Sugeno e Tsukmaoto fuzzy models and Layer 4 Layer 5 uses a

hybrid learning algorithm [26].


Layer 2
A1

IfM

Figure 3. Learning Vector Quantization Neural Network Structure Here, input layer consists of data provided. In the competitive layer, each competitive unit corresponds to a cluster and the output layer has the neurons equal to the no of classes. According to LVQ algorithm, (a) If <Z Wc

Figure 4. Adaptive Neuro-fuzzy Inference Systems The ANFIS can be trained by a hybrid learning algorithm [28]. In the forward pass the algorithm uses least- squares method to identify the consequent parameters while in the backward pass the errors are propagated backward and the premise parameters are updated by gradient descent [27]. Layer 1 is the input layer.

=Xi
(2)

w c (n + 1) = w c (n) + a n [x t - w c (n)]
where, 0 < a n < 1.
else

y(1) = x(1)
Layer 2 is the fuzzification layer.

(4)

Cc ^^x,
(3)

W c (n + 1) = W c (n) -a n [x t - w c (n)]

Where, {w^^ represent the set of Voronoi vectors,

Cw
2x(2) - ab
0,

if x(2) < a

2 b_ A2) b 2 (5) b 2 b if x(2) > a + 2

represent the class associated with the Voronoi vector, {xi}N=1 represent the set of input vectors, C x represent the class label of the input vector xi, and a n denotes the learning constant with the number of iterations n. (b) The other Voronoi vectors are not modified. LVQ network has an advantage over perceptron as they can classify any set of input vectors whereas perceptron can classify only linearly separable sets of input vectors. The only requirement is that the competitive layer must have enough neurons and each class must be assigned enough competitive neurons [23][24][25].

\1 -0,

3.3. Diagnosis using Inference System

Adaptive

Neuro-Fuzzy

y(2) =

The basic structure of the type of fuzzy inference system seen thus far is a model that maps input characteristics to input membership functions, input membership function to rules, rules to a set of output characteristics, output characteristics to output membership functions, and the output membership

Layer 3

JOURNAL OF COMPUTING, VOLUME 3, ISSUE 3, MARCH 2011, ISSN 2151-9617 HTTPS://SITES.GOOGLE.COM/SITE/JOURNALOFCOMPUTING/

Layer 3 is the inference layer.


(3 y ) = x) (3 x ) x2 x. . . x x,) i (3 (3

(6)

Layer 4 is the output membership layer

,( 4)

(4 )

(7)

(8)
WWW.JOURNALOFCOMPUTING.ORG

yr

x2 4 4 -

xu

Minimize (in w,b)

||l subject to (for any 1=1,2,3, ,n)

Layer 5 is the defuzzification layer. For, the datasets C1 and C2, the weighted average would be calculated as,
LI

ft frix - *)! 4. Experiments ai nd Results

(14)

xa

xb

+n +L

xa xb

xbL

xb

3.4. Diagnosis using Support Vector Machine (SVM)


Support Vector Machine (SVM) is based on the linear classifiers and on a general method known as "kernel tricks", which is based on the concepts of duality and kernels that allow algorithms to be formulated such that they could be systematically extended to non-linear cases. Support Vector Machine (SVM) models are similar to multilayer perceptron neural networks. Using a kernel function these are an alternative training method for polynomial, radial basis function and multi-layer perceptron classifiers. These classifiers are found to be helpful in solving a quadratic programming problem with linear constraints as compared to others. In SVM, let us assume the training data to be D, a set of n points of the form

The results are 6measured against the following diagnostic performance mea)sures: 1. True positive (TP): The number of positive cases correctly detected, 2. True negative (TN): The number of negative cases correctly detected, 3. 4. False positive (FP): The number of negative cases diagnosed as positive, and False negative (FN): The number of positive cases diagnosed as negative. The Wisconsin breast cancer diagnosis (WBCD) database is the result of ( the efforts made at the
8

S- K sc ft >taflffp,w{-U2

(9)

where, ci is either 1 or -1, indicating the class to which the point Xi belongs. Each Xi is a p-dimensional real vector. Any hyperplane can be written as the set of points X satisfying
H^K-fr" (10) We want to choose the W and b to maximize the margin. These hyperplanes can be described by the equation

University of Wisconsin Hospital for accurately diagnosing breast masses ba)sed solely on an FNA test. The database comprises diagnosis results of 699 patients beyond which 458 are benign and 241 are malignant [25]. Different neural network models were simulated for different parameter settings. The various performance measures are summarized and experimental results of breast cancer system using computational intelligence models are shown in table 1 and table 2 respectively. LVQ has emerged as the optimal network, which uses 10 hidden neurons and 0.01 learning rate. Table 1. Diagnostic performance measurement for Breast cancer. Absent Cancer Test Present Positive True Positive False Positive [TP] [FP] Negative False True Negative [FN] Total [TP+ FN] negative [TN] [TN + FP] Total [TP +FP]

& And

(11)

w - U- (12) If the training data are linearly separable then we can select the two hyperplanes of the margin in such a manner that there are no points between them and then try to maximize their distance
foralll<i<n (13) We can put this together to get the optimization problem:
7[(>F*Kc-&?&

[FN+TN]

[TP + FN+ TN + FP]

JOURNAL OF COMPUTING, VOLUME 3, ISSUE 3, MARCH 2011, ISSN 2151-9617 HTTPS://SITES.GOOGLE.COM/SITE/JOURNALOFCOMPUTING/

Sensitivity (Sens.) Specification (Spec.) Accuracy (Acc.)

TP / (TP + FN) TN / (TN + FP)

(TP + TN) / (TP + TN + FP + FN) False Positive Rate (FPR) FP / (TP+FP)


WWW.JOURNALOFCOMPUTING.ORG

False Negative Rate (FNR)

FN / (FN+TN)

Table 2. Performance comparisons of Computationals intelligence models


ANN Model BPA LVQ SVM ANFIS Sens. 80.00 95.00 90.00 88.00 Spec. 90.00 98.00 92.00 90.00 Acc. 85.00 96.50 91.00 89.00 FPR 20.00 05.00 10.00 12.00 FNR 10.00 02.00 08.00 10.00 TT (sec.) 2.00 57.00 05.00 10.00

Screening Mammograms Prior to Detection of Breast Cancer" IEEE journal of selected topics in signal processing, vol. 3, no. 1, February 2009, pp 46-52. [5] Al Mutaz M, Abdalla , Safaai Deris, Nazar Zaki and Doaa M. Ghoneim " Breast Cancer Detection Based on Statistical Textural Features Classification" 2008 IEEE , pp 728-730. [6] T.Z. Tan , C. Quek , G.S. Ng a, E.Y.K. Ng "A novel cognitive interpretation of breast cancer thermography with complementary learning fuzzy neural memory structure" Expert Systems with Applications 33 (2007) 652-666. [7] Shlomi Laufer and Boris Rubinsky "Tissue Characterization with an Electrical Spectroscopy SVM Classifier" IEEE transactions on biomedical engineering, vol. 56, no. 2, February 2009 pp 525528. [8] Taio C. S. Santos-Andrk and Anttinio C. Roque da Silva "A Neural Network Made of a Kohonen's SOM Coupled to a MLP Trained Via Backpropagation for the Diagnosis of Malignant Breast Cancer from Digital Mammograms" 1999 , IEEE pp 36473650. [9] Kenneth Revett, Florin Gorunescu, Marina Gorunescu, Elia El-Darzi and Marius Ene "A Breast Cancer Diagnosis System: A Combined Approach Using Rough Sets and Probabilistic Neural Networks ", EUROCON 2005, Serbia & Montenegro, Belgrade, November, 22-24, 2005,pp 1124-1127. [10] Xiangchun Xiong, Yangon Kim, Yuncheol Baek, Dae Wong Rhee, Soo-Hong Kim "Analysis of Breast Cancer Using Data Mining & Statistical Techniques" Proceedings of the Sixth International Conference on Software Engineering, Artificial Intelligence, Networking and Parallel/Distributed Computing and First ACIS International Workshop on Self-Assembling Wireless Networks (SNPD/SAWN'05) 2005 IEEE. [11] Emad Nabil, Amr Badr ,Ibrahim Farag, Mohamed Osama Khozium " A Hybrid Artificial Immune Genetic Algorithm with Fuzzy Rules for Breast Cancer Diagnosis "INFOS2008, March 2729, 2008 Cairo-Egypt, pp 31-40. [12] David B. Fogel, Eugene C. Wasson, Edward M. Boughton, Vincent W. Porto, and Peter J. Angeline "Linear and Neural Models for Classifying Breast Masses "IEEE transactions on medical imaging, vol. 17, no. 3, June 1998, pp 485-488. [13] Shakti K. Davis, Barry D. Van Veen, Susan C. Hagness, and Frederick Kelcz " Breast Tumor Characterization Based on Ultrawideband Microwave Backscatter" IEEE transactions on biomedical engineering, vol. 55, no. 1, january 2008, pp 237-246. [14] Lubomir Hadjiiski, Berkman Sahiner, Heang-Ping Chan, Nicholas Petrick and Mark Helvie " Classification of Malignant and Benign Masses Based on Hybrid ART2LDA Approach "IEEE transactions on medical imaging, vol. 18, no. 12, December 1999 pp1178-1187. [15] Ky Van Ha "Hierarchical Radial Basis Function Networks" 1998 EEE 1893 PP 1893-1898. [16] Yuehui Chen, Yan Wang, and Bo Yang " Evolving Hierarchical RBF Neural Networks for Breast Cancer Detection " ICONIP 2006, Part III, LNCS 4234, Springer-Verlag Berlin Heidelberg 2006 ,pp. 137-144, 2006. [17] Nuryanti Mohd Salleh , Harsa Amylia Mat Sakim and Nor Hayati Othman "Neural Networks to Evaluate Morphological Features for Breast Cells Classification" IJCSNS International Journal of Computer Science and Network Security, VOL.8 No.9, September 2008, pp 51-58. [18] Nuryanti Mohd Salleh , Harsa Amylia Mat Sakim and Nor Hayati Othman "Neural Networks to Evaluate Morphological Features for Breast Cells Classification" IJCSNS International Journal of Computer Science and Network Security, VOL.8 No.9, September 2008, pp 51-58.

5. Conclusions
Breast cancer is a major cause of deaths and predominantly affects women more than the age of 40. All the systems mentioned in the introduction were simulated on the breast cancer database adopted from the University of Wisconsin (UCI) Machine Learning Repository. The ultimate goal is to make a diagnostic system that would assist the doctors in diagnosis of the disease which would hence prove to be a very useful system considering the present scenario where diseases are on a hike and there is a lack of availability of specialized doctors. Such a system learns from the past data which is a collection of a lot of information in itself. The system tries to extract this hidden information to make a generalized system for the detection of the disease. It may be noted that though we had tried to build an effective system and obtain a accuracy of 100%, which is the ultimate goal of medical diagnosis but we were able to achieve a maximum accuracy of 96.50% using LVQ with 10 hidden neurons and 0.01 learning rate which surpasses the SVM whose accuracy is found out to be 91%. It can be concluded from the above result that LVQ should be used for prediction of the breast cancer as compared to other computational intelligence models.

Acknowledgement
The authors sincerely acknowledge Prof. S.G. Deshmukh, Director ABV-IIITM, Gwalior, India for encouraging and providing facilities to carry out this research work. This work is sponsored by ABV-IIITM Gwalior.

References
[1] http://www.breastcancer.org [2] Yuanjiao MA, Ziwu WANG, Jeffrey Lian LU, Gang WANG, Peng LI Tianxin MA, Yinfu XIE ,Zhijie ZHENG "Extracting Micro-calcification Clusters on Mammograms for Early Breast Cancer Detection" Proceedings of the 2006 IEEE International Conference on Information Acquisition August 20 - 23, 2006, Weihai, Shandong, China ,pp499-504. [3] Heng-Da Cheng, Yui Man Lui, and Rita I. Freimanis "A Novel Approach to Microcalcification Detection Using Fuzzy Logic Technique" IEEE transactions on medical imaging, vol. 17, no. 3, June 1998, pp442-450. [4] Mohammad Sameti, Rabab Kreidieh Ward, Jacqueline Morgan-Parkes and Branko Palcic "Image Feature Extraction in the Last
WWW.JOURNALOFCOMPUTING.ORG

JOURNAL OF COMPUTING, VOLUME 3, ISSUE 3, MARCH 2011, ISSN 2151-9617 HTTPS://SITES.GOOGLE.COM/SITE/JOURNALOFCOMPUTING/

[19] A. Rahmoun, S. Berrani, "A Genetic-Based Neuro-Fuzzy Generator: NEFGEN" 2001 IEEE pp 18-23 [20] Carlos Andres Pena-Reyes, Moshe Sipper "A fuzzy-genetic approach to breast cancer diagnosis "Artificial Intelligence in Medicine 17 (1999) 131-155. [21] C.A. Pena Reyes "Breast Cancer Diagnosis by Fuzzy CoCo" Coevolutionary Fuzzy Modeling, LNCS 3204, Springer-Verlag Berlin Heidelberg 2004 pp 71-87, 2004. [22] H. seker, m.Odeyato, D. petrovie, R.N.G.Naguib, C. Bartoli, L. Alasilo and G.V. Sherbet, " Prognostic comparison of statistical, neural and fuzzy methods of analysis of breast cancer image cytometric data "2001 proc. Of the 23rd Annual IEEE MBS Int. Conference October 25-28, Istanbul, Turkey, pp 3811-3814. [23] Shukla Anupam, Tiwari RituJanghel R.R.,and Kaur Prabhdeep, Diagnosis of Thyroid Disorders using Artificial Neural Networks on 2009 IEEE International Advance Computing Conference (IACC 2009),Patiala, India, 6-7March 2009,pp:2722-2726. [24] Janghel R.R , Shukla anupam,Tiwari Ritu and Pritesh Tiwari , "Clinical Decision support system for fetal Delivery using Artificial Neural Network"2009 (NISS) IEEE International Conference on New Trends in Information and Service Science, June 30 - July 2, 2009, Beijing, China,pp:1070-1075. [25] Janghel R.R , Shukla anupam and Tiwari Ritu " Decision Support System for Fetal Delivery using Soft Computing Techniques" (2nd ICIS), 2009 International Conference on Interaction Sciences: Information Technology, Culture and Human, IEEE CPS series ,24-26 November, 2009 - Seoul, Korea, pp: 1514-1519. [26] S.N. Sivanandan,S.N Deepa 2007, principle of soft computing, Wiley India private limited. [27] Abraham, A. 2001, Neuro-Fuzzy Systems: Stateof-the-Art Modeling Techniques. In Connectionist Models of Neurons, Learning Processes, and Artificial Intelligence, Springer, Berlin, J. Mira,A. Prieto (Eds.). [28] Jeen-Shing Wang and C. S. George Lee, "Self-Adaptive Neuro-Fuzzy Inference Systems for Classification Applications", IEEE Transactions on Fuzzy Systems, 2002. [29] Jaime S. Cardoso, Joaquim F. Pinto da Costa and Maria J. Cardoso," SVMs Applied to Objective Aesthetic Evaluation of Conservative Breast Cancer Treatment" Proceedings of International Joint Conference on Neural Networks, Montreal, Canada, July 31 - August 4, 2005,pp 2481-2486.

You might also like