Elsevier

Digital Signal Processing

Volume 22, Issue 1, January 2012, Pages 124-132
Digital Signal Processing

Texture-based analysis of clustered microcalcifications detected on mammograms

https://doi.org/10.1016/j.dsp.2011.09.004Get rights and content

Abstract

The need for early detection of breast cancer has led to establishing screening programs that generate large volumes of mammograms to be analyzed. These analysis are time consuming and labor intensive. Computerized analysis of mammograms has been suggested as “second opinion” or “pre-reader”.

In this paper, we suggest a texture-based computerized analysis clusters of microcalcifications detected on mammograms in order to classify them into benign and malignant types.

The test of the proposed system yielded a sensitivity of 100%, a specificity of 87.77% and a good classification rate of 89%; the area under the fitted ROC-curve using the MedCalc Statistical Software was 0.968.

Section snippets

Alain Tiedeu received his doctorate degree from the University of Yaoundé I, Cameroon, in 1995. He has been teaching electronics, digital signal processing, artificial neural networks, digital image processing, and related subjects at the National Advanced School of Engineering for the past decade. Professor Tiedeu has also served as reviewer, program committee member, and on editorial advisory board of a number of international conferences and journals (IEEE SITIS conference series, WSEAS

References (43)

  • A.F. Laine et al.

    Mammographic feature enhancement by multiscale analysis

    IEEE Trans. Medical Imaging

    (1994)
  • H. Li et al.

    Fractal modelling and segmentation for the enhancement of microcalcifications in digital mammograms

    IEEE Trans. Medical Imaging

    (1997)
  • H.D. Cheng et al.

    A novel approach to microcalcification detection using fuzzy logic technique

    IEEE Trans. Medical Imaging

    (1998)
  • J.M. Mossi, A. Albiol, Improving detection of clustered microcalcifications using morphological connected operators,...
  • B.C. Wallet et al.

    A method for detecting microcalcifications in digital mammograms

    J. Digital Imaging

    (1997)
  • S. Dippel et al.

    Multiscale contrast enhancement for radiographies: Laplacian pyramid versus fast wavelet transform

    IEEE Trans. Medical Imaging

    (2002)
  • W.J.H. Veldkamp et al.

    Normalization of local contrast in mammograms

    IEEE Trans. Medical Imaging

    (2000)
  • L. Zhang, W. Qian, R. Sankar, D. Song, R. Clark, A new false positive reduction method for MCCs detection in digital...
  • S. Yu et al.

    A CAD system for the automatic detection of clustered microcalcifications in digitized mammograms films

    IEEE Trans. Medical Imaging

    (2000)
  • S. Yu et al.

    Automatic detection of clustered microcalcifications in digitized mammograms films

    J. Electronic Imaging

    (1999)
  • H.S. Zadeh, S.P. Nezhad, F.R. Rad, Shape-based and texture-based feature extraction for classification of...
  • Cited by (34)

    • Multiscale connected chain topological modelling for microcalcification classification

      2019, Computers in Biology and Medicine
      Citation Excerpt :

      These are classified as malignant based on certain characteristics like size, shape, form, number, density, distribution pattern and cluster pattern [2]. Breast microcalcifications are small spots of calcium deposits which are represented as white specks in mammograms [4,5] as shown in Fig. 1. Though most detected microcalcifications are benign, the presence of fine and granular patterned microcalcifications could be an early indication of breast carcinoma requiring further investigation and potentially treatment [4].

    • Hybrid methods for feature extraction for breast masses classification

      2018, Egyptian Informatics Journal
      Citation Excerpt :

      To improve the efficiency and precision of a CAD system, it is important to extract the most discriminative features in an efficient way. Texture features have been commonly used in the analysis and interpretation of mammogram images [7,8]. Texture features based approaches use Gabor filters [9], Local Binary Pattern (LBP) [10,11], and Discreet Cosine Transform (DCT) to encode texture information of mammogram images [12].

    • Mammogram classification using sparse-ROI: A novel representation to arbitrary shaped masses

      2016, Expert Systems with Applications
      Citation Excerpt :

      As the earlier works in statistical matrices applicable to rectangular shape image only, two new algorithms are developed namely GLCM_Sparse-ROI and GLAM_Sparse-ROI. The features extracted from these are listed as (Tiedeu, Daul, Kentsop, Graebling, & Wolf, 2012; Mohanty, Senapati, & Lenka, 2013c): (1) ASM Homogenity; (2) Contrast; (3) Local homogenity; (4) Correlation; (5) Dissimilarity; (6) Entropy; (7) Sum of squares; (8) Interia; (9) Cluster shade; (10) Cluster prominence (11) Difference Entropy (12) Sum Entropy; (13) Sum average. As the SVM is the popular classification method, a Matlab function MultiSVM is used to classify the mammograms.

    • A new method of detecting micro-calcification clusters in mammograms using contourlet transform and non-linking simplified PCNN

      2016, Computer Methods and Programs in Biomedicine
      Citation Excerpt :

      A variety of techniques have been used in different steps. For mammogram enhancement, variety attempts have been done, such as improved histogram equalization [4], image enhancement based on wavelet fusion [5], automated lesion intensity enhance [6], modified multifractal analysis [7], etc.; in the segmentation step, many techniques have been suggested, such as multistable cellular neural networks, geodesic active contours (GAC) technique associated with anisotropic texture filtering [8], case-adaptive decision rule method [9], new scale-specific blob detection technique [10], etc.; in the third step, select true MCs by extracting a group of features of micro-calcifications like moment-based geometrical features [11], wavelet feature and Gabor feature [12] and so on. These aforementioned techniques make great contributions, however because the MCs detection faces different difficulties, the hybrid detection algorithms combining different theories seems more popular.

    View all citing articles on Scopus

    Alain Tiedeu received his doctorate degree from the University of Yaoundé I, Cameroon, in 1995. He has been teaching electronics, digital signal processing, artificial neural networks, digital image processing, and related subjects at the National Advanced School of Engineering for the past decade. Professor Tiedeu has also served as reviewer, program committee member, and on editorial advisory board of a number of international conferences and journals (IEEE SITIS conference series, WSEAS conference series, RPBME, etc.). A former regular associate member of the Abdus Salam International Centre for Theoretical Physics, his research interests include biomedical instrumentation and modelling, and medical signal and image processing and analysis.

    Christian Daul received the Ph.D. degree in computer vision from the Université Louis Pasteur (ULP), Strasbourg, France, in 1994. From 1990 to 1995, he was with the Laboratoire des Sciences de lʼImage, de lʼInformatique et dela Télédétection (LSIIT UMR 7005 CNRS/UdS) before joining the Institut of Industrial Mathematics (ITWM, Fraunhofer Institut), Kaiserslautern, Germany, where he was a member of the Image Processing Group. Since October 1999, he has been with the Centre de Recherche en Automatique de Nancy (CRAN UMR 7039 CNRS/Nancy University), Nancy, France, where he is currently working in the area of medical imaging (mammography, endoscopy, radiotherapy, and cardiology). His main research interests include image segmentation, data registration, and 3D data reconstruction. He is University Professor at the Institut National Polytechnique de Lorraine (EEIGM/INPL), Vandœuvre-Les-Nancy, France, where he is teaching in the signal processing field.

    Aude Kentsop is a holder of a Master degree in Computer Science from the University of Yaoundé I since 2004. She is currently on a Ph.D. program at the “Ecole de Technologie Supérieure”, Canada.

    Pierre Graebling received the Ph.D. degree in Image Processing from the Université Louis Pasteur (ULP), Strasbourg, France, in 1992. He was a Professor in computer science of the University of Strasbourg, Illkirch, France, where he was a member of the EAVR team of the Laboratoire des Sciences de lʼImage, de lʼInformatique et de la Télédétection (LSIIT). His research interests have included image processing, computer vision and medical imaging. In addition to the above topics, his teaching interests were programming, networks, and operating systems.

    Didier Wolf received the Ph.D. degree in Electrical Engineering from the Institut National Polytechnique de Lorraine, Nancy, France, in 1986. Currently, he is University Professor at the Institut National Polytechnique de Lorraine, where he is teaching in the signal processing field. Since 2005, he has been the Deputy Director of the Centre de Recherche en Automatique de Nancy (CRAN UMR 7039 CNRS/Nancy University), where he is the head of the biomedical engineering team. His main research interests include image processing, signal processing, and medical imaging techniques applied in the fields of cancerology and cardiology.

    1

    Professor Graebling passed away in February 2011 from a cancer. The authors dedicate this paper to him. It was a great pleasure working with him as a colleague and scientist. Those who were close to him enjoyed his capacity for dialog, his deep sense of humanity and humor and his courage in the face of trials.

    View full text