Effective features based on normal linear structures for detecting microcalcifications in mammograms

Abstract

Many features have been proposed for the detection of microcalcification clusters (MCCs) or classification of benign/malignant MCCs. However, most of them were designed based on the characteristics of MCC. In this paper, 16 features, which have been commonly adopted in many applications, are examined and six new features based on the linear structure are proposed. To evaluate the effectiveness of these six features, 800 suspicious regions detected from 320 full-field mammograms are equally divided into two parts for training and testing respectively. Experiments demonstrate that the area under the receiver operating characteristic (ROC) is increased from 0.86 to 0.89 after the new features are added into the set of feature selection. In the best feature sequence selected by the sequential floating forward search (SFFS) algorithm, the new proposed features take up the half number of features in the sequence

Effective recognition of MCCs in mammograms using an improved neural classifier

“…Unfortunately, such imbalance distribution is widely found for MCCs classification, as usually there are much more ( 44 times) benign samples than malignant ones in the training sets Wu et al, 2008). Therefore, the performance of the classifier may bias to the majority class and fails for correct detection of MCCs.…”

“…To detect suspicious MCC regions, optimal filtering using texture measurements is employed (Wu et al, 2006(Wu et al, , 2008. Firstly, some preprocessing is applied to remove the influence of background and several artifacts like white/black spots and scratches.…”

“…The features include local statistics and texture modeling (Grim et al, 2009), wavelets (Rashed et al, 2007;Soltanian-Zadeha et al, 2004;Heinlein et al, 2003;Sentelle et al, 2002), and morphological features (Kallergi, 2004), and RIOs can be determined via segmentation followed by local thresholding, spatial filtering and active contours (Cheng et al, 2003;Hassanien, 2007; Cheng et al, 2004;. From the segmented ROIs, MCCs can be detected using heuristics (Sentelle et al, 2002), fuzzy sets (Hassanien, 2007;Cheng et al, 2004), sub-image decomposition and filtering (Nakayama et al, 2006), and machine learning algorithms such as decision tree, neural network and support vector machine Sajda et al, 2002;Bocchi et al, 2004), where shape features such as linear structure is widely used (Zwiggelaar et al, 2004;Nakayama et al, 2006;Ge et al, 2006;Wu et al, 2008).…”

Breast cancer detection and classification in digital mammography based on Non-Subsampled Contourlet Transform (NSCT) and Super Resolution

“…Networks [2], [14], [20], [27], Nearest Neighbor [30], Fuzzy [25], [31] and Support Vector Machine (SVM) [3], [7], [13], [15], [30], [32]. Timp and Karssemeijer [1], developed CAD techniques to study interval changes between two consecutive mammographic screening rounds.…”

“…Then, the quality of mammographic images is improved. Several techniques have been proposed for pre-processing such as thresholding [12][13][14], region-based techniques [15][16][17] and edge detection techniques [18,19]. In the feature extraction stage, the features are extracted from mammographic images so that the system can correctly classify benign and malign lesions.…”