Identification of Masses and Microcalcifications in the Mammograms Based on Three Neural Networks: Comparison and Discussion

The incidence of breast cancer has been quickly raised in China recently. Mammography is regarded as the most reliable detection method of breast tumor and with the increase of cases, computer-aided diagnosis/detection (CAD) has been widely studied and applied to assist the radiologists. In the previous research, we had studied the detection algorithm of the microcalcifications (MCs) and the masses. And the identification on whether a focus is positive or negative is presented in this paper. In order to choose an effective classifier, three neural networks are used to identify the focuses respectively, and their performance is compared and then discussed. The experiments show that CMAC has very different performance for the training set and the testing set., ANFIS only performs well when the dimensionality of the features of the samples is comparatively low, while MLP is the most stable classifier, especially when the dimensionality of the features increases and the number of the samples is limited. And the final classification precision of the focuses is: for 120 mammograms of 60 cases, the true positive rate of the masses and the MCs are respectively 93.6% and 96.9%, and the false positive per image of them are 0.63 and 0.2.

[1]  Yianni Attikiouzel,et al.  Automatic pectoral muscle segmentation on mediolateral oblique view mammograms , 2004, IEEE Transactions on Medical Imaging.

[2]  Rangaraj M. Rangayyan,et al.  Analysis of asymmetry in mammograms via directional filtering with Gabor wavelets , 2001, IEEE Transactions on Medical Imaging.

[3]  Lihua Li,et al.  Detection and Classification of Microcalcifications Based on DWT and ANFIS , 2007, 2007 1st International Conference on Bioinformatics and Biomedical Engineering.

[4]  Huilong Duan,et al.  Segmentation of mass in mammograms using a novel intelligent algorithm , 2006, Int. J. Pattern Recognit. Artif. Intell..

[5]  Nikolas P. Galatsanos,et al.  A support vector machine approach for detection of microcalcifications , 2002, IEEE Transactions on Medical Imaging.

[6]  Ping Xu,et al.  Estimation of Breast Skin-Line Using Least Square Estimation and Watershed Segmentation , 2007, 2007 1st International Conference on Bioinformatics and Biomedical Engineering.

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

[8]  XU Wei-dong.  Study on model-based pectoral-muscle segment algorithmin mammograms , 2005 .

[9]  Y. Attikiouzel,et al.  Automatic pectoral muscle segmentation on mammograms by straight line estimation and cliff detection , 2001, The Seventh Australian and New Zealand Intelligent Information Systems Conference, 2001.

[10]  Yianni Attikiouzel,et al.  A simple method for automatically locating the nipple on mammograms , 1997, IEEE Transactions on Medical Imaging.

[11]  Robert M. Nishikawa,et al.  Relevance vector machine for automatic detection of clustered microcalcifications , 2005, IEEE Transactions on Medical Imaging.

[12]  Weidong Xu,et al.  Application of CMAC-Based Networks on Medical Image Classification , 2004, ISNN.

[13]  Robert M. Nishikawa,et al.  Computer-aided detection of clustered microcalcifications , 1992, [Proceedings] 1992 IEEE International Conference on Systems, Man, and Cybernetics.

[14]  C.J.S. deSilva,et al.  Spatially based application of the minimum cross-entropy thresholding algorithm to segment the pectoral muscle in mammograms , 2001, The Seventh Australian and New Zealand Intelligent Information Systems Conference, 2001.

[15]  Rangaraj M. Rangayyan,et al.  Detection of breast masses in mammograms by density slicing and texture flow-field analysis , 2001, IEEE Transactions on Medical Imaging.