Classification of breast mass abnormalities using denseness and architectural distortion

Abstract

This paper presents an electronic second opinion system for the classification of mass abnormalities in mammograms into benign and malignant categories. This system is designed to help radiologists to reduce the number of benign breast cancer biopsies. Once a mass abnormality is detected and marked on a mammogram by a radiologist, two textural features, named denseness and architectural distortion, are extracted from the marked area. The denseness feature provides a measure of radiographic denseness of the marked area, whereas the architectural distortion feature provides a measure of its irregularity. These features are then fed into a neural network classifier. Receiver operating characteristic (ROC) analysis was conducted to evaluate the system performance. The area under the ROC curve reached 0.90 for the DDSM database consisting of 404 biopsy proven masses. A sensitivity analysis was also performed to examine the robustness of the introduced texture features to variations in sizes of abnormality markings.

An Effective Method for Mammograph Image Retrieval

“…Therefore, we focused on the structures of the abnormal mammograms in this paper. Denseness and architectural distortion were introduced to describe texture feature in [8]. However, they were merely used to distinguish differences between benign and malignant masses, which did not adapt to more micromesh absolute classification.…”

“…Or else, the gradient becomes smaller and smaller by step [7]. It is obvious that the denseness mentioned in [8] couldn't describe this gradient transformation. In this paper, a new method is given.…”

Multidimensional Wavelet Analysis for Recognition of Lesions in Colposcopy Test

“…The analysis of temporal changes on the cervix surface has been performed using the metrics presented in (Baeg, Kehtarnavaz, 2002) to classify breast mass. These metrics are well correlated with colposcopy exam because features of lesions are similar to the breast mass (Tapia, Barreto and Altamirano, 2006).…”

“…Architectural distortion is associated with new details revealed by the acetic acid and it can be quantified using a gradient based measure like in (Baeg, Kehtarnavaz, 2002), because new details appears like small points, mosaics and edges. The equation used for its calculation is presented in (3).…”

On the use of local and global search paradigms for computer-aided diagnosis of breast cancer

“…Therefore, it is hard to make a fair comparison between performance of different CAD systems. (Chan et al, 1995) 168 Linear Discriminant Analysis (LDA) : 0.82 (Sahiner et al, 1996) 168 Convolution Neural Network (CNN) : 0.87 (Kinoshita et al, 1998) 92 Back Propagation Neural Network (BPNN) Acc: 0.81 (Hadjiiski et al, 1999) 348 Hybrid (Art2 and LDA) : 0.81 (Bruce & Adhami, 1999) 60 A simple Euclidian Metric Acc: 0.83 (Mudigonda et al, 2001) 56 LDA : 0.9 (Verma & Zakos, 2001) 58 BPNN Acc: 0.889 (Baeg & Kehtarnavaz, 2002) 404 ANN-MLP : 0.9 (De Santo et al, 2003) 102 ANN-MLP : 0.79 (Campanini et al, 2004) 512 Support Vector Machine (SVM) Sens: 0.80 (Guo et al, 2005) 40 Radial-Base Function (RBF) Neural Network, SVM Acc: 0.725 (Bellotti et al, 2006) 3369 ANN-MLP Sens: 0.80, : 0.783 (Timp et al, 2007) 465 SVM : 0.77 (Al Mutaz et al, 2011) 120 Artificial Neural Network (ANN)-Multi-Layer Perceptron (MLP) Sens: 0.916 Spec: 0.841 (Dinesh, 2011) 190 SVM Sens: 0.96 Spec: 0.97…”