Neural network fusion strategies for identifying breast masses

In this work, we introduce the perceptron average neural network fusion strategy and implemented a number of other fusion strategies to identify breast masses in mammograms as malignant or benign with both balanced and imbalanced input features. We numerically compare various fixed and trained fusion rules, i.e., the majority vote, simple average, weighted average, and perceptron average, when applying them to a binary statistical pattern recognition problem. To judge from the experimental results, the weighted average approach outperforms the other fusion strategies with balanced input features, while the perceptron average is superior and achieves the goals with lowest standard deviation with imbalanced ensembles. We concretely analyze the results of above fusion strategies, state the advantages of fusing the component networks, and provide our particular broad sense perspective about information fusion in neural networks.

[1]  E A Sickles,et al.  Screening mammography in community practice: positive predictive value of abnormal findings and yield of follow-up diagnostic procedures. , 1995, AJR. American journal of roentgenology.

[2]  Dimitris N. Metaxas,et al.  Optimal feature combination for automated segmentation of prostatic adenocarcinoma from high resolution MRI , 2003, Proceedings of the 25th Annual International Conference of the IEEE Engineering in Medicine and Biology Society (IEEE Cat. No.03CH37439).

[3]  Jiri Matas,et al.  On Combining Classifiers , 1998, IEEE Trans. Pattern Anal. Mach. Intell..

[4]  Jesús Cid-Sueiro,et al.  Cost functions to estimate a posteriori probabilities in multiclass problems , 1999, IEEE Trans. Neural Networks.

[5]  Rangaraj M. Rangayyan,et al.  Gradient and texture analysis for the classification of mammographic masses , 2000, IEEE Transactions on Medical Imaging.

[6]  Kagan Tumer,et al.  Structural adaptation and generalization in supervised feed-forward networks , 1994 .

[7]  Leo Breiman,et al.  Bagging Predictors , 1996, Machine Learning.

[8]  Yoav Freund,et al.  Experiments with a New Boosting Algorithm , 1996, ICML.

[9]  Maryellen L. Giger.  Computer-aided diagnosis in digital mammography , 1993, Proceedings of the 15th Annual International Conference of the IEEE Engineering in Medicine and Biology Societ.

[10]  Kagan Tumer,et al.  Analysis of decision boundaries in linearly combined neural classifiers , 1996, Pattern Recognit..

[11]  Fabio Roli,et al.  Design of Multiple Classifier Systems , 2002 .

[12]  George Cybenko,et al.  Approximation by superpositions of a sigmoidal function , 1989, Math. Control. Signals Syst..

[13]  Martin A. Riedmiller,et al.  A direct adaptive method for faster backpropagation learning: the RPROP algorithm , 1993, IEEE International Conference on Neural Networks.

[14]  T.,et al.  Training Feedforward Networks with the Marquardt Algorithm , 2004 .

[15]  J. Cid-Sueiro,et al.  Neural posterior probabilities for microcalcification detection in breast cancer diagnoses , 2003, First International IEEE EMBS Conference on Neural Engineering, 2003. Conference Proceedings..

[16]  Alicia Samuels,et al.  Cancer Statistics, 2003 , 2003, CA: a cancer journal for clinicians.

[17]  Eros Gian Alessandro Pasero,et al.  Multi-layer perceptron ensembles for pattern recognition: some experiments , 1994, Proceedings of 1994 IEEE International Conference on Neural Networks (ICNN'94).

[18]  Lars Kai Hansen,et al.  Neural Network Ensembles , 1990, IEEE Trans. Pattern Anal. Mach. Intell..

[19]  Kurt Hornik,et al.  Multilayer feedforward networks are universal approximators , 1989, Neural Networks.

[20]  S. Feig.  Role and evaluation of mammography and other imaging methods for breast cancer detection, diagnosis, and staging. , 1999, Seminars in nuclear medicine.

[21]  Yunfeng Wu,et al.  Fusing output information in neural networks: ensemble performs better , 2003, Proceedings of the 25th Annual International Conference of the IEEE Engineering in Medicine and Biology Society (IEEE Cat. No.03CH37439).

[22]  Bruce W. Schmeiser,et al.  Improving model accuracy using optimal linear combinations of trained neural networks , 1995, IEEE Trans. Neural Networks.

[23]  M. Møller.  A Scaled Conjugate Gradient Algorithm for Fast Supervised Learning , 1990 .

[24]  Mohammad Bagher Menhaj,et al.  Training feedforward networks with the Marquardt algorithm , 1994, IEEE Trans. Neural Networks.

[25]  Rangaraj M. Rangayyan,et al.  Classification of tumors and masses in mammograms using neural networks with shape and texture features , 2003, Proceedings of the 25th Annual International Conference of the IEEE Engineering in Medicine and Biology Society (IEEE Cat. No.03CH37439).

[26]  R. Rangayyan,et al.  Boundary modelling and shape analysis methods for classification of mammographic masses , 2000, Medical and Biological Engineering and Computing.

[27]  Martin Fodslette Møller,et al.  A scaled conjugate gradient algorithm for fast supervised learning , 1993, Neural Networks.

[28]  Josef Kittler,et al.  Fixed and trained combiners for fusion of imbalanced pattern classifiers , 2002, Proceedings of the Fifth International Conference on Information Fusion. FUSION 2002. (IEEE Cat.No.02EX5997).

[29]  James T. Kwok,et al.  Efficient cross-validation for feedforward neural networks , 1995, Proceedings of ICNN'95 - International Conference on Neural Networks.

[30]  D.R. Hush,et al.  Progress in supervised neural networks , 1993, IEEE Signal Processing Magazine.

[31]  Mohamed S. Kamel,et al.  Decision fusion in neural network ensembles , 2001, IJCNN'01. International Joint Conference on Neural Networks. Proceedings (Cat. No.01CH37222).

[32]  Ludmila I. Kuncheva,et al.  A Theoretical Study on Six Classifier Fusion Strategies , 2002, IEEE Trans. Pattern Anal. Mach. Intell..

[33]  Athina P. Petropulu,et al.  ROC analysis of ultrasound tissue characterization classifiers for breast cancer diagnosis , 2003, IEEE Transactions on Medical Imaging.