Volume 23, Issue 2 (June & July 2020)                   J Arak Uni Med Sci 2020, 23(2): 246-263 | Back to browse issues page

XML Persian Abstract Print

Download citation:
BibTeX | RIS | EndNote | Medlars | ProCite | Reference Manager | RefWorks
Send citation to:

Safdarian N, Yousefian Dezfoulinejad S. Mammographic Image Processing for Classification of Breast Cancer Masses by Using Support Vector Machine Method and Grasshopper Optimization Algorithm. J Arak Uni Med Sci 2020; 23 (2) :246-263
URL: http://jams.arakmu.ac.ir/article-1-6190-en.html
1- Young Researchers and Elite Club, Tabriz Branch, Islamic Azad University, Tabriz, Iran. , naser.Safdarian@yahoo.com
2- Department of Biomedical Engineering, Faculty of Engineering, Dezfoul Branch, Islamic Azad University, Dezfoul, Iran.
Full-Text [PDF 9110 kb]   (3590 Downloads)     |   Abstract (HTML)  (3028 Views)
Full-Text:   (12382 Views)


Breast cancer is a type of cancer that, due to the abnormal growth of cells, causes a lump in the breast tissue. According to the World Health Organization (WHO), this cancer affects 2.1 million women each year, and also causes the highest number of cancer deaths among women [1]. Mammography X-ray imaging is one of the most common methods used by radiologists to diagnose and screen for breast cancer and to determine the presence of lumps and cysts. In mammography images, very fine calcareous particles are usually seen as noisy particles, and the masses have very low light intensity, making it difficult for radiologists and physicians to detect. Given that accurate and timely diagnosis of cancerous masses, as well as its various types, is of particular importance in the health of individuals in society, the difficulty of diagnosing breast cancer masses, which is mainly associated with diagnostic errors, should be addressed by researchers. In this study, we introduce a new and automatic method to diagnose and detect breast cancer masses with high accuracy. For this purpose, after pre-processing and detecting the borders of the cancerous masses from mammography images, a number of features were extracted from the detected masses and in the end, the feature matrix was applied to the Support Vector Machine (SVM) classifier input. 

Materials and Methods

The images used in this study were collected from the Digital Database for Screening Mammography (DDSM) database [17]. First, for pre-processing of raw images, a 3×3 median filter was applied on digital mammography images to remove noise using MATLAB software. Then, the threshold method was used to extract the cancerous masses. Since the mass edge extracted by the threshold method had inward direction, the brightness of the pixels around the edge was expanded towards the center by using the Dilation operator. After detecting the area of breast cancer masses, we extracted 19 structural features from this area by MATLAB software. Finally, using SVM parameter optimization method by Grasshopper Optimization Algorithm (GOA), as well as using 4-fold cross validation method, data were divided into two categories of benign and cancer.


The values of accuracy, sensitivity, and specificity (along with their variances) resulting from the use of data (benign and cancer) classified by the SVM method using three kernel functions of Linear, Radial Basis Function (RBF) and Polynomial were presented in tables. The final results after applying the GOA were also shown in a separate table. The used training data was 85% and 15% of the data were considered as test data. In 4-fold crossvalidation method, the number of programs executed per kernel function was 100 times. The best results of accuracy, sensitivity and specificity indicators (features) for using RBF kernel function in SVM classifier (before process) were obtained 97%, 100% and 96%, respectively. For linear function after optimization of SVM parameters by GOA, it was obtained 100% for all accuracy, sensitivity, and specificity indicators, which shows the high accuracy of the proposed method. The average values of accuracy, sensitivity and specificity indices for all three SVM kernel functions after applying the optimization algorithm were 95.83, 100 and 94.81%, respectively.


Some advantages of this study include a large number of features extracted from masses detected from mammography images, the use of GOA to more accurately determine the type of breast tissue cancer mass, and high speed and accuracy of the proposed algorithm. The boundaries of cancerous tumors were extracted with high accuracy, and finally the classification was performed using simple morphological features. To our knowledge, no study has previously used the optimization methods in the final classification stage. So, it can be said that this the first study that use the GOA to optimize the kernel parameters of different SVM classifiers. This can be the advent of new methods in improving various classification processes in a variety of medical diagnoses. After the detection and diagnosis of breast cancer masses that was performed with high accuracy in this study, according to the morphological and simple features of cancer masses, classification operation was performed well and with high accuracy. The results of this study show the higher performance of the proposed method compared to other methods used in previous related studies [6-16]. 

Ethical Considerations

Compliance with ethical guidelines

Images from DDSM database were used in this research, all images are open access in this database.


This research did not receive any grant from funding agencies in the public, commercial, or non-profit sectors.

Authors' contributions

Conceptualization, research, methodology: Naser Safdarian; Data collection, resources, writing-original draft: Shadi Yousefian Dezfoulinejad; writing - review & editing: Naser Safdarian.

Conflicts of interest

The authors declared no conflict of interests.


The authors would like to thank the Young Researchers and Elite Club of Islamic Azad University, Tabriz branch.



1.World Health Organization. Cancer: Early diagnosis and screening [Internet]. 2020 [Retrived 6 Aug 2020]. Available at: https://www.who.int/cancer/prevention/diagnosis-screening/en

2.Center of Disease Control. [Report of registered cancer cases (Persian)]. Tehran: Center of Disease Control; 2004. 

3.Baines CJ, McFarlane DV, Miller AB. The role of the reference radiologist. Estimates of inter-observer agreement and potential delay in cancer detection in the national breast screening study. Investigative radiology. 1990; 25(9):971-6. [DOI:10.1097/00004424-199009000-00002] [PMID] 

4.Wallis MG, Walsh MT, Lee JR. A review of false negative mammography in a symptomatic population. Clin Radiol. 1991; 44(1):13-5. [DOI:10.1016/S0009-9260(05)80218-1] 

5.Behnam H, Zakeri F, Gifani P, Torkashvand P, Shalbaf A, [Ultrasound Imaging Processing (Persian)]. Tehran: Ishraqiya Publishing; 2011.

6.Nick Ravan Shalmani A, Karami Mohammadi M. [Diagnosis of breast cancer masses in computer aided mammography images (Persian)]. The 3rd International Conference on Recent Innovations in Electrical and Computer Engineering, 9 September 2016; Tehran, Iran. 

7.Abbaspour Kazerouni I, Haddad Nia J. [Introducing a precise intelligent system for mammographic image separation based on density of tissues and masses (Persian)]. Iran J Breast Dis. 2013; 6(1).

8.Naseri Noroozani S, Shayegan MA. [Clinical stage detection of breast cancer patients using tnm system and ant colony algorithm (Persian)]. Iran J Breast Dis. 2018; 11(3):56-70.

9.Jabbari H, Bigdeli N, Khadem A. [A new hybrid approach to segmentation and diagnosis of tumors in breast mammography images (Persian)]. Iran J Breast Dis. 2016; 9(3):14-24.

10.Tavakkolah P, Safabakhsh R., [A new approach to classifying and classifying breast cancer masses (Persian)] Third Information and Knowledge Technology Conference. 6-8 December 2007: Tehran; Iran.

11.Sheikhpour Ro, Sheikhpour Ra. [Diagnosis of breast cancer using nonparametric estimation of nuclear methods-based probability density (Persian)]. Razi J Med Sci. 2016; 23(144):30-40.

12.Pezeshki H, Rastgarpour M, Sharifi A, Yazdani S. Extraction of spiculated parts of mammogram tumors to improve accuracy of classification. Multimed Tools Appl. 2019; 78:1-25. [DOI:10.1007/s11042-019-7185-4] 

13.Mughal B, Sharif M, Muhammad N, Saba T. A novel classification scheme to decline the mortality rate among women due to breast tumor. Microsc Res Tech. 2018; 81(2):171-80. [DOI:10.1002/jemt.22961] [PMID] 

14.Torres W, Oseas A, Sousa A, Silva FA. Functional Diversity applied to the false positive reduction in breast tissues based on digital mammography. 2018 IEEE Symposium on Computers and Communications (ISCC), Natal. 2018; 25:1120-25. [DOI:10.1109/ISCC.2018.8538658] 

15.Mohamed BA, Salem NM. Automatic classification of masses from digital mammograms. 2018, 35th National Radio Science Conference (NRSC). 2018 Mar 20; 495-502. IEEE. [DOI:10.1109/NRSC.2018.8354408] 

16.Pérez M, Benalcázar ME, Tusa E, Rivas W, Conci A. Mammogram classification using back-propagation neural networks and texture feature descriptors. 2017 IEEE Second Ecuador Technical Chapters Meeting (ETCM). 2017 Oct 16; 1-6. [DOI:10.1109/ETCM.2017.8247515] 

17.Digital Database for Screening Mammography. Available at: http://marathon.csee.usf.edu/Mammography/Database.html

18.Vadivel A, Surendiran B. A fuzzy rule-based approach for characterization of mammogram masses into BI-RADS shape categories. Comput Biol Med. 2013; 43(4):259-67. [DOI:10.1016/j.compbiomed.2013.01.004] [PMID] 

19.Huang T, Yang GJ, Tang G. A fast two-dimensional median filtering algorithm. IEEE Trans Acoust Speech Signal Process. 1979; 27(1):13-8.  [DOI:10.1109/TASSP.1979.1163188] 

20.Massodi P, Safdarian N, Kalantar B. [Detection and classification of breast cancer using feature extraction from mammography images by image processing technique and neural network (Persian)]. International Conference on Non-Linear System & Optimization in Computer & Electrical Engineering. May 2015: Dubai; United States of Emirates.

21.Fradkin D, Muchnik I. Support vector machines for classification. DIMACS series in discrete mathematics and theoretical computer science. 2006; 70:13-20.

22.Saremi S, Mirjalili S, Lewis A. Grasshopper optimisation algorithm: Theory and application. Adv Eng Softw. 2017;105:30-47. [DOI:10.1016/j.advengsoft.2017.01.004] 

23.Safdarian N, Hedyezadeh M. Detection and classification of breast cancer in mammography images using pattern recognition methods. Multidiscip Cancer Investig. 2019; 3(4):13-24 URL: http://mcijournal.com/article-1-229-en.html [DOI:10.30699/acadpub.mci.3.4.13] 

24.Ghaffari H, Mostashari M, Mahmoodi M. [Designing a system for detection of pulmonary nodules in lung ct images using support vector machine classifier (Persian)]. J Health Biomed Informatics, Med Informatics Res Center. 2017; 3(4):300-9.

25.Biniaz A, Abbasi A, Shamsi M. [A comprehensive overview of segmentation of medical images of magnetic resonance in the human brain: methods, clinical applications, advantages and disadvantages (Persian)]. Majlis Electrical Engineering Conference. 2012: New Majlis City: Iran.

Type of Study: Original Atricle | Subject: Obstetrics & Gynocology
Received: 2019/11/13 | Accepted: 2020/01/21

Add your comments about this article : Your username or Email:

Send email to the article author

Rights and permissions
Creative Commons License This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License.

© 2024 CC BY-NC 4.0 | Journal of Arak University of Medical Sciences

Designed & Developed by : Yektaweb