Comparison of SVM and neural network classifiers in automatic detection of clustered microcalcifications in digitized mammograms

This paper presents a computer-aided diagnosis (CAD) system for automatic detection of clustered microcalcifications (MCs) in digitized mammograms. The proposed system consists of two main steps. First, potential microcalcification pixels in the mammograms are segmented out by using 4 mixed features consisting of two wavelet features and two gray level statistical features and then the potential microcalcification pixels are labeled into potential individual microcalcification objects by their spatial connectivity. Second, MCs are detected by using a set of 17 features extracted from the potential individual microcalcification objects. The classifier which is used in the first step is a multilayer feedforward neural network (NN) classifier but for the second step we have used three different classifier which are multilayer feedforward neural network (NN), SVM with polynomial kernel and SVM with Gaussian RBF kernel and the result of each classifier is obtained separately. The method is applied to a database of 40 mammograms (Nijmegen database) containing 105 clusters of MCs. A free-response operating characteristics (FROC) curve is used to evaluate the performance of CAD system with each classifier. Results show that the proposed system gives quite satisfactory detection performance. In particular, 89.55% mean true positive detection rate is achieved at the cost of 0.782, 1 and 0.95 false positive per image for neural network, support vector machine (SVM) with polynomial kernel and SVM with Gaussian RBF kernel classifiers, respectively.