Segmentation of microcalcification in X-ray mammograms using entropy thresholding
We describe a new algorithm for microcalcification segmentation in mammographic X-ray images. The algorithm detects microcalcifications in two steps. First, it removes background tissue with a multiscale morphological operation. Then, it applies entropy t
- PDF / 639,739 Bytes
- 6 Pages / 481.89 x 685.984 pts Page_size
- 68 Downloads / 179 Views
'CARS/Springer. All rights reserved.
671
Segmentation of microcalcification in X-ray mammograms using entropy thresholding Moti Melloul and Leo Joskowicz School of Computer Science and Engineering The Hebrew University of Jerusalem Jerusalem 91904, ISRAEL E-mail: [email protected]@cs.huji.ac.il
Abstract We describe a new algorithm for microcalcification segmentation in mammographic xray images. The algorithm detects microcalcifications in two steps. First, it removes background tissue with a multiscale morphological operation. Then, it applies entropy thresholding based on a 3-dimensional co-occurrence matrix. Unlike existing methods, ours is fully automatic, parameter-free, and independent of local statistics. To test its efficacy, we applied it to images from the Mammographic Image Analysis Society database and analyzed the results with the assistance ofa clinician. We obtained detection rates of93.75% of true positives, 6.25% offalse positives, and 2% offalse negatives. Keywords: X-ray mammograms, microcalcification segmentation, entropy thresholding.
1. Introduction Most early breast cancer can be diagnosed by detecting microcalcification clusters in mammographic X-ray images. The clusters appear as groups of small, bright particles with arbitrary shapes. Detecting microcalcifications is difficult because they are embedded in a non-homogeneous background. Many missed radiologist diagnoses can be attributed to human factors such as subjective or varying decision criteria, distraction by other image features, large number of images to be inspected, or simple oversight. Therefore, there is strong motivation to develop reliable and effective methods for automatic microcalcifications detection. While many methods for microcalcification segmentation have been developed in the past ten years, they either require manual thresholds adjustment or depend on local statistics to compute those thresholds. This paper presents a new fully automatic, parameter-free, and local statistics independent algorithm for microcalcification segmentation in mammographic X-ray images. For a detailed description of the method, see [I].
2. Previous work Strickland and Hahn [2] describe a method that uses multiscale matched filters with wavelet transforms for enhancing and detecting calcifications. Nishikawa et at [3] use a difference technique to enhance microcalcifications. First, it extracts potential microcalcifications with global thresholding based on an erosion operator and local
CARS 2002 - H. U. Lemke, M W. Vannier; K. Inamura, A.G. Farman, K. Doi & J.H.c. Reiber (Editors) 'CARS/Springer. All rights reserved.
672
adaptive thresholding. False positives are then eliminated by texture analysis, and the remaining candidates are grouped with a non-linear clustering algorithm. Cheng et al. [3] propose a method based on fuzzy logic, which consists of image fuzzification, enhancement, irrelevant structure removal segmentation, and reconstruction. Chan et al. [4] investigate a convolution neural network based approach that is
Data Loading...
