A comparison of two methods for the segmentation of masses in the digital mammograms
Introduction
World widely the breast cancer is the leading and the second most fatal disease in women [1]. It is one of the most deadly cancers among middle-edged women. There have been a host of consistent efforts to mitigate the malady. Successful treatment is a key to reduce the high death rate. An early detection of the breast cancer leads to the successful treatment while that at the advanced stages is usually impossible to cure. X-ray mammography is the most important modality for detecting the early-stage breast cancer. Currently, X-ray mammography is the clinical Gold Standard for the detection of breast cancer. It is a well understood and standardized procedure. It works fairly well in the postmenopausal women and is inexpensive. Consequently, the breast cancer is an intensely researched field [2], [3], [4], [5], [6], [7].
Now coming to the main theme of this paper, it may be said that Image segmentation aims at partitioning the image into physically meaningful regions. The ultimate goal is that this partition identifies the objects of interest in the image. One such application of segmentation is in the medical image analysis for the purpose of diagnosis. Morphological segmentation techniques are quite sought after in this respect [8], [9].
The breast region is identified by the presence of higher gray values than that of the non-breast region. Therefore, thresholding is often employed for the segmentation of the breast region [10]. Masses are more difficult to detect than micro calcifications because the features of a mass bear semblance to those of the normal breast parenchyma. A mass is demarcated as a space-occupying lesion seen in more than one projection and is usually characterized by its shape and margin. A mass with regular shape has a higher probability of being benign whereas a mass with an irregular shape has a high probability of being malignant [11], [12].
A mammogram mainly contains two regions: the exposed breast region and the unexposed non-breast region. It is necessary to first identify the breast region for the reduction of the subsequent processing and then to remove the non-exposed breast region. Bick et al. [10] have explored a segmentation method for the breast region based on the morphological gradient calculation and the modified global histogram analysis [10]. Ball and Bruce [2] present an automated mammographic computer aided diagnosis system to detect and segments spicules. Mendez et al. [13] describe an automatic algorithm that computes the gradient of the gray levels. Wirth and Stapinski [14] make use of the snakes and fuzzy approach [15] for the segmentation. This paper makes a comparison of the level set method with the morphological marker controlled watershed approach for segmentation.
The paper is organized as follows. The proposed methodology is outlined in Section 2. Section 3 presents the results of the experimentation along with a validation by an expert radiologist. Conclusions are drawn in Section 4.
Section snippets
Data collection
Only mammograms that show positive for the malignant mass are collected for this study. The total number of cases is 17. The breast mammogram database is procured from the Mammography Image Analysis Society (MIAS) and NCR MRI Diagnosis Center, NIT, Faridabad, India. A sample mammogram of the collected database is shown in Fig. 1.
Results
The proposed algorithms are tested on 17 mammograms containing malignant masses. Expert-segmented data in all the images are provided in Table 3. All the 17 images are semi-automatically segmented and then validated from the expert-segmented ones.
Two novel segmentation approaches are applied on the mammogram images for the detection of region of interest (ROI) which includes both masses and the pectoral muscles. In the mammograms, masses are assumed to be distinctive regions that are relatively
Conclusions
Two novel semi-automated approaches are presented for the tumor image segmentation which overcomes the accuracy and sensitivity limitations of the current solutions.
As mammograms images are simple but susceptible to noise the preprocessing of image is very essential. The proposed algorithms seem to work well on the mammograms as seen from the exact boundary of the abnormal growth or lesions thus demonstrating their comparative edge over other methods.
The marker controlled watershed segmentation
Acknowledgements
Authors acknowledge the retrieval of MIAS database from the Internet for the experiments of this paper and wish to thank NCR MRI Diagnosis Center, NIT, Faridabad, India, for permitting us to use the databases.
Rash Bihari Dubey was born in India on 10 November 1961. He received the MSc physics with specialization in electronics in 1984 from Agra University Agra, India and MTech in instrumentation from R.E.C. Kurukshetra, India in 1989. He is pursuing the PhD degree as a part-time studentship on the topic “Computer Aided Diagnosis of Malignancy in Brain Tumor and Breast Tumor”. He is at present an asstt professor in the Department of Instrumentation and Control Engineering at Apeejay College of
References (20)
- et al.
Cancer incidence and morality in Europe
Ann Oncol
(2005) - et al.
Automatic segmentation of digitized mammograms
Acad Radiol
(1995) - et al.
Automatic detection of breast border and nipple in digital mammograms
Comput Methods Programs Biomed
(1996) - American Cancer Society. Breast cancer facts and figures 2005–2006; 2006. p....
- et al.
Digital mammographic computer aided diagnosis using adaptive level set segmentation
EMBS Proc IEEE
(2007) - et al.
Co-occurrence analysis for discovery of novel breast cancer pathology patterns
IEEE Trans Inform Technol Biomed
(2006) - et al.
Breast skin-line estimation and breast segmentation in mammograms using fast marching method
Int J Biol Med Sci
(2008) - et al.
Classification of breast cancer malignancy using cytological images of fine needle aspiration biopsies
Int J Appl Math Comput Sci
(2008) - et al.
A comparison of three different methods for classification of breast cancer data
- et al.
Visually significant dynamics for watershed segmentation
IEEE Int Conf Pattern Recognit
(2006)
Cited by (38)
Breast mass detection in digital mammography based on anchor-free architecture
2021, Computer Methods and Programs in BiomedicineCitation Excerpt :Although the traditional BMass detection method has certain limitations, it has made some progress [13–16]. For example, based on threshold segmentation [12], region growing [17,18], watershed [18,19], edge detection [20,21]. Specifically, Dubey R B et al. used a horizontal image set and a watershed two traditional image segmentation algorithms to segment the BMass, achieving an accuracy of 82% [19].
A fast and automated segmentation method for detection of masses using folded kernel based fuzzy c-means clustering algorithm
2019, Applied Soft Computing JournalAn image segmentation method based on network clustering model
2018, Physica A: Statistical Mechanics and its ApplicationsCitation Excerpt :From Fig. 5, it can be seen that although the suspicious region is slightly different from the ground truth, the proposed method segments the region of tumor successfully. Researchers have proposed many different kinds of segmentation evaluation methods [34–37]. As the ground truth of images used in this paper is known, several evaluation methods are adopted to compare the ground truth with the segmentation result obtained by algorithms.
Breast mass contour segmentation algorithm in digital mammograms
2013, Computer Methods and Programs in BiomedicineCitation Excerpt :Proposed system achieves varying precision and recall values around 80% and 85%. Dubey et al. [15] measure performance of level-set and watershed segmentation methods on mass segmentation. They found that Watershed segmentation algorithm produces better results than level-set segmentation approach with respect to relative error measure.
Pectoral muscle segmentation: A review
2013, Computer Methods and Programs in BiomedicineA pyramid-based mass detection scheme based on texture analysis and neural classifier for digital mammograms
2014, Biomedical Engineering - Applications, Basis and Communications
Rash Bihari Dubey was born in India on 10 November 1961. He received the MSc physics with specialization in electronics in 1984 from Agra University Agra, India and MTech in instrumentation from R.E.C. Kurukshetra, India in 1989. He is pursuing the PhD degree as a part-time studentship on the topic “Computer Aided Diagnosis of Malignancy in Brain Tumor and Breast Tumor”. He is at present an asstt professor in the Department of Instrumentation and Control Engineering at Apeejay College of Engineering, Sohna, Gurgaon, India. His research interest are in the areas of digital signal processing, digital image processing, biomedical signal analysis, medical imaging and industrial real time applications.
Madasu Hanmandlu received the BE degree in electrical engineering from Osmania University, Hyderabad, India, in 1973, the MTech degree in power systems from R.E.C. Warangal, Jawaharlal Nehru Technological University (JNTU), India, in 1976, and the PhD degree in control systems from Indian Institute of Technology, Delhi, India, in 1981. From 1980 to 1982, he was a senior scientific officer in Applied Systems Research Program (ASRP) of the Department of Electrical Engineering, IIT Delhi. He joined the EE department as a lecturer in 1982 and became assistant professor in 1990, an associate professor in 1995 and finally a professor in 1997. He was with Machine Vision Group, City University, London, from April to November, 1988, and Robotics Research Group, Oxford University, Oxford from March to June, 1993, as part of the Indo-UK research collaboration. He was a visiting professor with the Faculty of Engineering (FOE), Multimedia University, Malaysia from March 2001 to March 2003. He worked in the areas of power systems, control, robotics and computer vision, before shifting to fuzzy theory. His current research interests mainly include fuzzy modeling for dynamic systems and applications of fuzzy logic to image processing, document processing, medical imaging, multimodal biometrics, surveillance and intelligent control. He has authored a book on computer graphics in 2005 under PBP publications and also has well over 165 publications in both conferences and journals to his credit. He has guided 13 PhDs and 80 MTech students. He has handled several sponsored projects. He is presently an associate editor of both Pattern Recognition Journal and IEEE Transactions on Fuzzy Systems and a reviewer to other journals such as Pattern Recognition Letters, IEEE Transactions on Image Processing and Systems, Man and Cybernetics. He is a senior member of IEEE and is listed in Reference Asia; Asia's who's who of Men and Women of achievement; 5000 Personalities of the World (1998), American Biographical Institute.
Sushil Kumar Gupta was born in India on 21 May 1966. He received the BE in electrical engineering in 1990 from M.N.N.I.T. Allahabad, India, and the ME in Inf. Sci. & Eng. from M.N.N.I.T. Allahabad, India in 1994 and the PhD in power system engineering from M.D.U., Rohtak, India in 2002. He is at present professor in the Department of Electrical Engineering at DCRUST, Murthal, India. He is as a coordinator for MTech electrical engineering (I&C), MTech electrical engineering (Power Systems), MTech power & energy management. His research interests are in the areas of restructuring of electric power systems, power system control, SCADA & energy management, power system dynamics & FACTS, computer application to power systems and medical imaging.