Breast Cancer Diagnosis: Analyzing Texture of Tissue Surrounding Microcalcifications

Abstract

The current study investigates texture properties of the tissue surrounding microcalcification (MC) clusters on mammograms for breast cancer diagnosis. The case sample analyzed consists of 85 dense mammographic images, originating from the Digital Database for Screening Mammography. Mammograms analyzed contain 100 subtle MC clusters (46 benign and 54 malignant). The tissue surrounding MCs is defined on original and wavelet decomposed images, based on a redundant discrete wavelet transform. Gray-level texture and wavelet coefficient texture features at three decomposition levels are extracted from surrounding tissue regions of interest (ST-ROIs). Specifically, gray-level first-order statistics, gray-level cooccurrence matrices features, and Laws' texture energy measures are extracted from original image ST-ROIs. Wavelet coefficient first-order statistics and wavelet coefficient cooccurrence matrices features are extracted from subimages ST-ROIs. The ability of each feature set in differentiating malignant from benign tissue is investigated using a probabilistic neural network. Classification outputs of most discriminating feature sets are combined using a majority voting rule. The proposed combined scheme achieved an area under receiver operating characteristic curve ( A(z)) of 0.989. Results suggest that MCs' ST texture analysis can contribute to computer-aided diagnosis of breast cancer.

Analysis of perceived similarity between pairs of microcalcification clusters in mammograms

“…9,12,[23][24][25][26][27][28][29] These nine features can be categorized into three groups: (1) image features describing individual MCs (Feature Group 1), including the standard deviation of the image contrast values of the MCs, and the maximum and the standard deviation of the sizes of the MCs, 9,12,23,24 (2) spatial clustering features of the MCs (Feature Group 2), including the number of MCs in a cluster, the area of the cluster, and the compactness of cluster, 12,24,25 and (3) texture-based features (Feature Group 3), including the energy, contrast, and correlation derived from the gray-level cooccurrence matrices (GLCM). [26][27][28][29] For completeness, the detailed definitions of these features are given in Appendix.…”

“…The GLCM is a well-established robust tool for extracting second texture information in an image. 29 It represents the distribution of joint probability of occurrence of a pair of gray-level values separated by a given displacement d. In other words, it calculates how often a pixel with gray-level value i occurs adjacent to a pixel with the gray-level value j. In this study, the GLCM is obtained by setting d = [10,0], that is, the distance of 10 pixels in the horizontal direction is used.…”

Mammogram Image Segmentation Using Rough Clustering

“…The texture features were derived for each subregion from an averaged gray-level co-occurrence matrix (GLCM). Karahaliou et al [30] Gray-level texture and wavelet coefficient texture features at three decomposition levels are extracted from surrounding tissue regions of interest.…”

Texture Analysis Methods for Medical Image Characterisation

“…This traditional approach has been used extensively to describe different image textures by unique features and has found application in many disparate fields such as: discrimination of terrain from aerial photographs (Conners & Harlow, 1980); in vitro classification of tissue from intravascular ultrasound (Nailon, 1997); identification of prion protein distribution in cases of CreutzfeldJakob disease (CJD) (Nailon & Ironside, 2000); classification of pulmonary emphysema from lung on high-resolution CT images (Uppaluri et al, 1997;Xu et al, 2004;Xu et al, 2006); and identifying normal and cancerous pathology (Karahaliou et al, 2008, Zhou et al, 2007Yu et a., 2009). Higher-order approaches have been used to localise thrombotic tissue in the aorta (Podda, 2005) and to determine if functional vascular information found in dynamic MR sequences exists on anatomical MR sequences (Winzenrieth, 2006).…”