An automated mammogram classification system using modified support vector machine

Figures & data

Table 1 Textural features extracted from the dataset

S/N	Features	Mathematical Expression
1.	*IMC	$f_1=\frac{HXY-HXY1}{max(HX,HY)}$
2.	Contrast	$f_2=\sum _{m,n=1}^{N_g}|m-n{|}^{2}P(m,n)$
3.	Correlation	$f_3=\sum _{m,n=1}^{N_g}\frac{P(m,n)-{\mu }_x{\mu }_y^{}}{{\sigma }_x{\sigma }_y}$
4.	Cluster Prominence	$f_4=\sum _{m,n=1}^{N_g}{(m-{\mu }_x+n-{\mu }_y)}^{4}P(m,n)$
5.	Cluster Shade	$f_5=\sum _{m,n=1}^{N_g}{(m-{\mu }_x+n-{\mu }_y)}^{3}P(m,n)$
6.	Dissimilarity	$f_6=\sum _{m,n=1}^{N_g}|m-n|P(m,n)$
7.	Energy	$f_7=\sum _{m,n=1}^{N_g}(P(m,n){)}^2$
8.	Entropy	$f_8=-\sum _{m,n=1}^{N_g}P(m,n)log(P(m,n))$
9.	*IMC2	$f_9=\sqrt{(1-exp(-2(HXY2-HXY)))}$
10.	Difference Variance	$f_{10}=\sum _{l=0}^{N_g-1}{l}^2{P}_{x-y}(l)$
11.	Variance	$f_{11}=\sum _{m,n=1}^{N_g}{\left(1-\mu \right)}^{2}P(m,n)$
12.	Sum Average	$f_{12}=\sum _{l=2}^{2N_g}l..P_{x+y}(l)$
13.	Sum Variance	$f_{13}=\sum _{l=2}^{2N_g}(l-f_8{)}^2{P}_{x+y}(l)$
14.	Difference Entropy	$f_{14}=-\sum _{l=0}^{N_g-1}{P}_{x-y}(l)log(P_{x-y}(l))$
15.	Homogeneity	$f_{15}=\sum _{m,n=1}^{N_g}\frac{1}{1+{(m-n)}^2}P(m,n)$

Note: *IMC₁ and IMC₂ are information measure sof correlation 1 and 2, respectively.

Figure 1 Two-level classification approach using support vector machine (SVM).

Figure 2 Mammogram enhancement using Contrast Limited Adaptive Histogram Equalization (CLAHE).

Figure 3 Region of interest (ROI) extraction.

Figure 4 Gray level co-occurrence matrix (GLCM) features computed at D=1.

Figure 5 Gray level co-occurrence matrix (GLCM) features computed at D=2.

Figure 6 Weighted gray level co-occurrence matrix (GLCM) features computed at distances 1 and 2.

Figure 7 Overall average gray level co-occurrence matrix (GLCM) features computed.

Figure 8 First-stage of classification: (A) normal mammogram classification; (B) abnormal mammogram classification.

Figure 9 Second-stage classification: (A) benign; (B) malignant.

Table 2 The system’s confusion matrix

Expert’s classification	System classification		Total
	Benign	Malignant
Benign	TN=21	FP=2	23
Malignant	FN=1	TP=17	18
Total	22	19

Abbreviations: TN = True Negative; FN = False Negative; FP = False Positive and TP = True Positive.

Table 3 Results from proposed technique and existing works

S/N	Authors	Classification category	Sensitivity (%)	Specificity (%)	Accuracy (%)
1.	Lothe et al^Citation19	NA	92.30	62.50	86.84
2.	Domínguez and Nandi^Citation20	BM	55	85	PPV 0.71 NPV 0.75
3.	Rejani and Selvi^Citation21	NA	88.75	–	–
4.	Moayedi^Citation22	NA	95.8	99.0	96.60%
5.	Dheeba and Tamil^Citation23	NA	–	–	86.1%
6.	Kavitha and Thyagharajan^Citation24	BM	100	96	98.00
7.	Zhang et al^Citation25	BM	94.85%	78.20	–
8.	Kamra et al^Citation26	BM	71.43	97.22	93.02
9.	Rouhie et al^Citation27	BM	85.41	91.89	88.65
10.	Damasceno et al^Citation28	BM	99.41	99.84	99.73
11.	Kaur et al^Citation29	NA	–	–	96.90
	Proposed method	NA (BM)	(100) 94.4	(100) 91.3	(100) 92.68

Abbreviations: NA, MIAS classification into normal and abnormal mammograms; BM, MIAS classification into benign and malignant Mammograms; MIAS, Mammographic Image Analysis Society; NPV, negative predictive value; PPV, positive predictive value.

Savita L., Rupali T, Almas S, Prapti DD.Detection and classification of breast mass using support vector machine. IOSR J Comput Eng. 2011;1–6.

 Google Scholar

Domínguez AR, Nandi AK. Toward breast cancer diagnosis based on automated segmentation of masses in mammograms. Pattern Recognit. 2009;42:1138–1148. doi:10.1016/j.patcog.2008.08.006

 Web of Science ®Google Scholar

Rejani YIA, Selvi ST. Early detection of breast cancer using SVM classifier technique. Int J Comput Sci Eng. 2009;1(3):127–130.

 Google Scholar

Moayedi F, Azimifar Z, Boostani R, Katebi S. Contourlet-based mammography mass classification using the SVM family. Comput Biol Med. 2010;40(4):373–383. doi:10.1016/j.compbiomed.2009.12.00620181330

 PubMed Web of Science ®Google Scholar

Dheeba J, Tamil S. Classification of malignant and benign microcalcification using SVM classifier. 2011 International Conference on Emerging Trends in Electrical and Computer Technology 2011:686–690.

 Google Scholar

Kavitha S, Thyagharajan KK. Features Based Mammogram Image Classification Using Weighted Feature Support Vector Machine. International Conference on Computing and Communication Systems. Berlin (Heidelberg): Springer; 2011:320–329.

 Google Scholar

Zhang E, Wang F, Li Y, Bai X. Automatic detection of microcalcifications using mathematical morphology and a support vector machine. Biomed Mater Eng. 2014;24(1):53–59.2. doi:10.3233/BME-130783

 PubMed Web of Science ®Google Scholar

Kamra A, Jain VK, Singh S, Mittal S. Characterization of architectural distortion in mammograms based on texture analysis using support vector machine classifier with clinical evaluation. J Digit Imaging. 2016;29(1):104–114.26138756

 PubMed Web of Science ®Google Scholar

Rouhi R, Jafari M, Kasaei S, Keshavarzian P. Benign and malignant breast tumors classification based on region growing and CNN segmentation. Expert Syst Appl. 2015;42(3):990–1002. doi:10.1016/j.eswa.2014.09.020.

 Web of Science ®Google Scholar

Damasceno E, Oseas A, Filho DC, et al. Method of differentiation of benign and malignant masses in digital mammograms using texture analysis based on phylogenetic diversity. Comput Electr Eng. 2018:210–222. doi:10.1016/j.compeleceng.2018.03.038

 Web of Science ®Google Scholar

Kaur P, Singh G, Kaur P. Intellectual detection and validation of automated mammogram breast cancer images by multi-class SVM using deep learning classification. Inf Med Unlocked. 2019;1–19. doi:10.1016/j.imu.2019.01.001

 Google Scholar

Data availability

Raw mammograms used in this work are publicly available at mini-MIAS: http://peipa.essex.ac.uk/pix/mias/.