International Journal of Emerging Technology and Advanced Engineering

Website: www.ijetae.com (E-ISSN 2250-2459, Scopus Indexed, ISO 9001:2008 Certified Journal, Volume 12, Issue 03, March 2022)
Manuscript Received: 01 February 2022, Received in Revised form: 05 March 2022, Accepted: 09 March 2022 DOI: 10.46338/ijetae0322_13

A Review and Computational Analysis of Breast Cancer Using

Different Machine Learning Techniques
Varsha Nemade1, Sunil Pathak2, Ashutosh Kumar Dubey3, Deepti Barhate4
1,2
Amity School of Engineering and Technology, Department of Computer Science & Engineering, Amity University Rajasthan,
Jaipur, India
3
Chitkara University School of Engineering and Technology, Chitkara University, Himachal Pradesh, India
4
SVKM’s NMIMS MPSTME Shirpur, Maharashtra, India
Abstract—Breast cancer is found to be prime cause of Out of these we have considered total of 69 papers for
women death in the current era. It is mainly due to the late the review and analysis (Figure. 1). It has been selected
detection of the disease. Artificial intelligence and machine based on methods related mostly with artificial intelligence
learning plays an important role in breast cancer image techniques, machine learning and Convolution Neural
classification and lesion detection. This paper focuses on the
Network (CNN). The main methods considered from the
study and analysis of breast image classification, feature
extraction and feature selection. It also covers the role of artificial intelligence and machine learning domains. CNN
Convolution Neural Network (CNN) architecture with the is used, for the feature’s extraction adaptively according to
impact of transfer learning. Furthermore, this survey also targeted output. In this paper image preprocessing
discusses different database repositories which are publicly techniques, feature extraction and selection techniques
available for breast cancer with their features. Overall, this were discussed mostly. Then dataset has also been
study and analysis helps the researcher working in this area discussed from the related work.
for the further investigation. This paper mainly covers the following aspects:
Keywords— Breast cancer, Artificial intelligence, Machine • To review and analyze the methods in the direction of
learning, CNN breast cancer.
• To explore the dataset where experimentation can be
I. INTRODUCTION performed and validated.
According to World Health Organization (WHO), breast 60
cancer is fifth leading cause of cancer mortality worldwide
[1]. According to Indian Council of Medical Research- 50
50
Population Based Cancer Registries (ICMR-PBCR) data, 40
Count

breast cancer is the commonest cancer among women in 30

urban registries in India [2]. There are broadly two type of
20
breast cancer according to their origin breast: Ductal cancer
begin in the ducts that carry milk to nipple, Lobular Cancer 10 16
1 2
begins in the glands that produce breast milk. Malignant 0
and benign are also types of breast cancer depend on types before 2005 2006-2010 2011-2015 2016-2020
of cells in tumor. Various medical imaging techniques use
Years
for breast cancer analysis such as X-ray, ultrasound, MRI
scan, PET scan, CT scan [3−6]. These images are used to
identify abnormalities, masses but only this is not sufficient FIGURE 1. PAPERS CONSIDERED YEAR WISE UP TO 2020
for the accurate prediction. It is also important to identify
whether the cell is cancerous or not, for these different II. RELATED WORK
types of images are used called biopsy images [7−12]. In this section study and discussion were performed
The main aim is to analyse the breast cancer detection based on dataset and the breast cancer detection methods.
methods with the major challenges and advantages. For the Mammogram, ultrasound or biopsy images are used for
survey and analysis, we have considered 237 total papers. breast cancer diagnosis.

111
 International Journal of Emerging Technology and Advanced Engineering
Website: www.ijetae.com (E-ISSN 2250-2459, Scopus Indexed, ISO 9001:2008 Certified Journal, Volume 12, Issue 03, March 2022)
Some publicly available Datasets are described in For classification of breast cancer using ultrasound
Table1. There are two views acquired for Breast such as: a images instead of conventional CAD, Adaptive CAD
Cranio-Caudal (CC) and a Mediolateral Oblique View (A-CAD) system, in this morphological feature represents
(MLO). Different types of features like intensities, shapes, the size of tumor on this adaptive filtering is applied and it
textures can be extracted from images, they are useful for create two datasets on which separately logistic regression
analysis. classification is applied [34]. Casti et.al suggested the
For doing breast cancer diagnosis using images different approach for analysis of asymmetry between right and left
method are used such as conventional machine learning side of mammogram of patient, which helps to find out the
and CNN. From images first feature are extracted then presence of tumor in mammogram in early stages [35]. On
classification of images performed using different ultrasound images, by applying Super resolution algorithm
techniques such as Support Vector Machine (SVM), for texture analysis after that selects ROI and then apply
Decision Tree (DT), Random Forest (RF), k -Nearest feature extraction and finally classification using random
Neighbor (k-NN). Images used for diseases prediction are forest improves the performance [17].
of different types. It is needed to pre-process because they Deep learning and machine learning are the part of
may contain noise and inconsistencies. By pre-processing it artificial intelligence. In Traditional machine learning many
helps to improve performance of the model. Table 2 shows steps needs to perform classification and the performance
different pre-processing techniques applied on breast also depends on the selected features. Deep Learning
cancer images. allows automated learning of features which are used for
In image classification, feature extraction is important classification [43−48]. Pan and Yang has described how
step. In conventional machine learning, textural features deep learning shows remarkable performance in image
give the information about the intensity of image as well as classification or image modalities task. Most cases
structural and dimensional information of image. In cancer performed this by weight transferring [49]. Sharif et al.
detection, identification of nucleus, area and shape of described the approach in which features generated by
nucleus is the important information; this can be obtained activation of second network, these features send to any
by using structural features of image. Not all extracted classifier [50]. In some cases, weights are fine tuned. Over
features are always important. Table 3 shows different fitting occurs when fine tuning applied on target dataset
feature extraction techniques for breast cancer images. For which is too small than the capacity of network. If over
selection of features: Principal Component Analysis (PCA), fitting is not the problem then perform fine tune on every
Linear Discriminant Analysis (LDA), and layer [51]. Deciding how many layers to transfer depend on
Multidimensional Scaling (MDS) are mainly used [15]. the closeness of two tasks also the imaging modalities [52].
Different algorithms are used for classification of images Following section describe how different CNN
such as k-NN, SVM, Naïve Bayes, Decision tree and architectures are useful in breast cancer diagnosis using
random forest. Already lots of research has been done on mammograms and ultrasound images. Table 5 shows deep
breast mass, microcalcification classification and detection learning method applied on different breast cancer datasets
mentioned in literature [25−31]. Detection of masses is and their performance measure.
very challenging task as they are varying in size [32]. Table Histopathalogy is the category of images uses for
4 shows different dataset and classification techniques diagnosis of breast cancer. Albarqouni etal [68] proposed
applied on that with their performance. Khan et al. new CNN which learning from different CNN model.
described different approaches for mass classification in They added new layer called aggregation layer in existing
mammogram using Gabor feature extraction and they used CNN to perform mitosis detection. By using this AggNet
Successive Enhancement Learning based weighted Support model got AUC 0.86. Spanhol et al. [69] used AlexNet
Vector Machine (SELwSVM) for classification [33]. network for breast cancer classification using pathological
images in dataset BreakHis, they got higher accuracy than
traditional machine learning methods. Bayramoglu et al.
[14] also apply deep learning on BreakHis dataset and got
approximately 83% classification accuracy.

112
 International Journal of Emerging Technology and Advanced Engineering
Website: www.ijetae.com (E-ISSN 2250-2459, Scopus Indexed, ISO 9001:2008 Certified Journal, Volume 12, Issue 03, March 2022)
TABLE 1.
PUBLICLY AVAILABLE BREAST CANCER DATASET

Dataset Type Image Size of Distribution Strengths Weakness

Format images
Digital Database for Mammogram LJPEG 3118×5001 Normal: 914 -Large and widely used. Old format of
Screening Benign: 870 -It contains MLO View images.
Mammography Malignant: 695 and CC view
(DDSM) [6]
Mammographic Image Mammogram PGM 1024×1024 Normal: 207 Small but widely used -Very small.
Analysis Society Benign: 69 for research -Has only one
(MIAS) [6] Malignant: 56 MLO view
INBreast [13] Mammogram DIACOM different Normal: 67 -Several types of lesions Limited size
Benign: 220 were included,
Malignant: 49 -Accurate contours made
by specialists
Breast Cancer Digital Mammogram TIF 720×1168 Different Standard format Limited size
Repository (BCDR) DIACOM different according category
[8]
Break-his dataset [14] biopsy PNG 700×460 Benign:2480 Large and widely used Data imbalance
Malignant:5429

TABLE 2.
PRE-PROCESSING TECHNIQUES ON BREAST CANCER IMAGES

Author Pre-processing method Use

Jen et al. [5] -Global Equalization Transformation, -To normalize the range of contrast variation
-Mean filter, among different images,
-Binarization -To remove short-tailed noise
-Identify the breast region and remove the non-
breast region
The major aim of this approach is the
identification of the contrast from different set of
images which can be helpful in the proper breast
region identification.
Rouhi et al. [7] -Histogram equalization -The intensity of image pixels is stretched to
-Median filter extend the contrast.
-To remove salt and paper noise
This approach is helpful in the prediction of the
image pixels intensity. It is also helpful in the
removal of the noise.
Arevalo et al. [8] -Global Contrast Normalization (GCN) For adjusting pixel values of image
-Local Contrast Normalization (LCN) This approach is helpful in the normalization of
the pixel intensity values to enhance the
corresponding associated image pixels.
Carneiro et al. [9] Otsu segmentation To select bounding box for segmentation
This segmentation will be helpful in the threshold
condition optimization for the extraction and
connotations for the correlating values as it will
be helpful in the same set of segments.
Djebbar et al. [10] Multi-threshold peripheral equalization To remove the effect of breast compression
Gnanasekaran et al. [11] Contrast limited adaptive histogram Enhancement of local contrast
equalization (CLAHE)

113
 International Journal of Emerging Technology and Advanced Engineering
Website: www.ijetae.com (E-ISSN 2250-2459, Scopus Indexed, ISO 9001:2008 Certified Journal, Volume 12, Issue 03, March 2022)
TABLE 3.
DIFFERENT FEATURE EXTRACTION TECHNIQUES ON BREAST CANCER IMAGES
Author Feature Extraction Method
Beura et al. [6] Discrete Orthonormal S Transform (DOST). This method provides feature
extraction with the help of orthonormal transformation.
Bruno et al. [12] -Fast Discrete curvelet transform (FDCT) via wrapping method
-Local binary pattern (LBP)
This method provides transformation with the wrapping methods.
Raghavendra et al. [16] Gabor Wavelet
Abdel-Nasser et al. [17] -LBP
-Gray Level Co-occurrence Matrix (GLCM)
Muduli et al. [18] Lifting Wavelet Transform (LWT)
Song et al. [19] Histogram of Oriented Gradient (HOG)
Muramatsu et al. [20] RLTP, GLCM, Discrete Wavelet transform (DWT)
Diz et al. [13] GLCM
Zhang et al. [21] Difference of Gaussian (DoG) filter
de Lima et al. [22] Multiresolution wavelet
Vijayarajeswari et al. [23] Hough Transform
Jadoon et al. [24] Discrete Curvelet Transform (DCT)

TABLE 4.
DIFFERENT FEATURE EXTRACTION TECHNIQUES ON BREAST CANCER IMAGES

Author Dataset used Classification Techniques used Performance Measure with results
Beura et al. [6] MIAS Classification: AdaBoost with RF Area under the ROC Curve (AUC) 0.99
DDSM AUC: 0.99
Dong et al. [36] DDSM RF AUC: 0.95
MIAS AUC: 0.94
Diz et al. [13] INBreast Naïve Bayes, SVM, k-NN and DT 3a-scenario using k-NN classifier (77.0%
global accuracy)
Random Forest classifier at 3b-scenario with a
global accuracy of 66.0%
Zhang et al. [21] Private RF false positive locations with 40%
dataset
de Lima et al. [22] IRMA Multilayer Perceptron (MLP), Accuracy 94.11%
SVM, ELM
Bilal et al. [37] QT SVM Accuracy 90%
ultrasound
Muramatsu et al. [20] Private ANN, SVM, RF Max AUC 0.90
dataset
Peng et al. [38] MIAS ANN Accuracy 96%
Bruno et al. [12] DDSM Polynomial classifier AUC
BCDR-FMR 0.91 ,1,1for (normal malignant)
BCDRDMR 0.94,1,1(for benign malignant)
Raghavendra et al. [16] DDSM k-NN 98.69%
Jalalian et al. [39] CTLM MLPNN 95.2%
Li et al [40] DDSM SVM 99.66%
Cordeiro et al. [41] Mini Mias Fuzzy grow cut (segmentation) p-value
0.976
Sahran et al. [42] Breast UKM Ensemble frame work Accuracy 90.8 ± 5.0
SVM
Vijayarajeswari et al. [23] MIAS SVM Accuracy 94%

114
 International Journal of Emerging Technology and Advanced Engineering
Website: www.ijetae.com (E-ISSN 2250-2459, Scopus Indexed, ISO 9001:2008 Certified Journal, Volume 12, Issue 03, March 2022)
TABLE 5.
COMPARISON USING DIFFERENT METHODOLOGY USING DEEP LEARNING AND TRANSFER LEARNING
Author Dataset used Methodology Performance Measure with
results
Jiao et al. [53] DDSM Deep features from different layers and SVM Accuracy 96.7
Arevalo et al. [8] BCDR Features extraction through CNN and SVM AUC: 0.826
Huynh et al. [54] University of Chicago AlexNet and SVM AUC: 0.81
Medical Center
Sun et al. [15] In house full-field Using CNN AUC: 0.8818
digital mammography
(FFDM) image
database
Samala et al. [55] digitized screen-film Transfer learning in Deep CNN Single task transfer
mammograms (SFMs) 0.78±0.02,
and digital multi-task transfer learning
mammograms (DMs), 0.82±0.02, lesion-based
DDSM performance0.76±0.01
view-based performance
0.79±0.01,
Antropova et al [56] FFDM Pre-trained CNN using VGG19, CADx, fusion AUC: 0.86
Ultrasound classifier, SVM AUC:0.90
DCE-MRI AUC: 0.89
Dhungel et al. [57] INBreast Pre-train CNN using LeNet, RF AUC: 0.91 ± 0.12
Qiu et al. [58] FFDM CNN with logistic regression AUC0.696±0.044 (for 1-fold),
AUC 0.802±0.037(for 2fold),
AUC 0.836±0.036(for 3-Fold),
and AUC 0.822±0.035(for 4fold)
AUC for entire dataset
0.790±0.019.
Jadoon et al. [24] Image Retrieval in CNN, SVM AUC for CNN-WT: 0.846,
Medical Applications AUC for CNN-CT: 0.855
(IRMA)
Zhang et al. [59] Private dataset AlexNet AUC for 2D mammogram
0.7274
AUC for 3D mammogram
0.6632
Jiao et al. [60] DDSM Parasitic metric learning network Accuracy 97.4%
MIAS Accuracy 96.7%
Chougrad, et al. [61] DDSM Pertained CNN suing VGG-16, ResNet50, Accuracy 97.35%
INbreast InceptionV3 95.50%
BCDR 96.67%
Al-masni et al. [62] DDSM CNN(YOLO) Accuracy 97%
Ribli et al. [63] IN Breast, DDSM, Faster (Region based CNN) R-CNN AUC:0.95
dataset from the
Semmelweis University
in Budapest
Wang et al. [64] BCDR RNN AUC:0.89
Yu et al. [65] BCDR GoogleLeNet, AlexNet, CNN2 and CNN3, AUC :0.88
SVM AUC:0.83
AUC:0.82
Cai et al. [66] Private dataset. Pre-trained using AlexNet Accuracy:0.8768
1. Kaur et al. Mini MIAS Features extracted using CNN, K-mean Accuracy
[67] clustering, MSVM/decision tree Normal:95%,
Benign:94%
Malignant:98%,

115
 International Journal of Emerging Technology and Advanced Engineering
Website: www.ijetae.com (E-ISSN 2250-2459, Scopus Indexed, ISO 9001:2008 Certified Journal, Volume 12, Issue 03, March 2022)

III. DISCUSSIONS This paper helps researcher to get the details about the
different dataset, techniques applied and their results, also
This paper helps to identify the recent research
provides possible solutions for improving the performance
development in breast cancer using different image
for breast cancer at one place.
modalities. Breast cancer analysis can be done using
machine learning and deep learning techniques such CNN, REFERENCES
Ensemble learning. [1] Torre LA, Islami F, Siegel RL, Ward EM, Jemal A. Global cancer in
Performance: In Breast cancer diagnosis using machine women: burden and trends. Cancer Epidemiology and Prevention
learning, features from images are extracted by using Biomarkers. 2017; 26(4):444-57.
different feature extraction techniques discussed this paper. [2] Agarwal G, Ramakant P. Breast cancer care in India: the current
CNN extract features automatically. Different CNN scenario and the challenges for the future. Breast Care. 2008;
3(1):21-7.
architecture used for breast cancer analysis also discussed
[3] Lomboy KE, Hernandez RM. A comparative performance of breast
in paper and their performance also compared. Transfer cancer classification using hyper-parameterized machine learning
learning is also one of the best options for improving models. International Journal of Advanced Technology and
performance and to save time of training, by using this we Engineering Exploration. 2021 Sep 1;8(82):1080.
can use pretrained weights and apply it on CNN model. [4] Kumar R, Srivastava R, Srivastava S. Detection and classification of
Overfitting problem: Performance of the model also cancer from microscopic biopsy images using clinically significant
and biologically interpretable features. Journal of Medical
depends on the size of dataset. Image dataset available for Engineering. 2015;2015.
breast diagnosis are very small, so to avoid over fitting [5] Jen CC, Yu SS. Automatic detection of abnormal mammograms in
problem data augmentation can be used. mammographic images. Expert Systems with Applications. 2015;
Ensemble Learning: As different classifier shows 42(6):3048-55.
different performance on same datasets. In this case [6] Beura S, Majhi B, Dash R, Roy S. Classification of mammogram
Ensemble learning can be used. using two‐dimensional discrete orthonormal S‐transform for breast
cancer detection. Healthcare Technology Letters. 2015; 2(2):46-51.
Based on the study and analysis the following
[7] Rouhi R, Jafari M, Kasaei S, Keshavarzian P. Benign and malignant
observations were noted: breast tumors classification based on region growing and CNN
1. The need of hybrid framework for the detection and segmentation. Expert Systems with Applications. 2015; 42(3):990-
prediction with different parameters. 1002.
2. Machine learning algorithms are found to be suitable in [8] Arevalo J, González FA, Ramos-Pollán R, Oliveira JL, Lopez MA.
Representation learning for mammography mass lesion classification
the detection of the breast cancer with different level of with convolutional neural networks. Computer Methods and
computational capabilities. Programs in Biomedicine. 2016; 127:248-57.
3. CNN models are also efficient in the evaluation and [9] Carneiro G, Nascimento J, Bradley AP. Automated analysis of
performance accuracy. unregistered multi-view mammograms with deep learning. IEEE
4. The major challenge is the data with the complete Transactions on Medical Imaging. 2017; 36(11):2355-65.
annotation for the experimentation. [10] Djebbar K, Mimi M, Berradja K, Taleb-Ahmed A. Deep
convolutional neural networks for detection and classification of
tumors in mammograms. In International Conference on Image and
IV. CONCLUSIONS Signal Processing and Their Applications 2019 (pp. 1-7). IEEE.
Breast cancer is one of the main causes of death in [11] Gnanasekaran VS, Joypaul S, Sundaram PM, Chairman DD. Deep
learning algorithm for breast masses classification in mammograms.
women. It is important to reduce the mortality rate in IET Image Processing. 2020; 14(12):2860-8.
women by studying breast cancer in detail, for this [12] Bruno DO, Do Nascimento MZ, Ramos RP, Batista VR, Neves LA,
technology helps a lot. Using different image modalities Martins AS. LBP operators on curvelet coefficients as an algorithm
and different CNN models breast cancer can be detected to describe texture in breast cancer tissues. Expert Systems with
earlier. It describes different image pre-processing, feature Applications. 2016; 55:329-40.
extraction techniques applied on breast images; also [13] Diz J, Marreiros G, Freitas A. Using data mining techniques to
support breast cancer diagnosis. In new contributions in information
describe the feature selection and classification methods. It systems and technologies 2015 (pp. 689-700). Springer, Cham.
also shows that by using deep learning and transfer learning [14] Bayramoglu N, Kannala J, Heikkilä J. Deep learning for
performance of breast classification is improved. magnification independent breast cancer histopathology image
classification. In international conference on pattern recognition
2016 (pp. 2440-2445). IEEE.

116
 International Journal of Emerging Technology and Advanced Engineering
Website: www.ijetae.com (E-ISSN 2250-2459, Scopus Indexed, ISO 9001:2008 Certified Journal, Volume 12, Issue 03, March 2022)
[15] Sun W, Tseng TL, Zhang J, Qian W. Enhancing deep convolutional [30] Sert E, Ertekin S, Halici U. Ensemble of convolutional neural
neural network scheme for breast cancer diagnosis with unlabeled networks for classification of breast microcalcification from
data. Computerized Medical Imaging and Graphics. 2017; 57:4-9. mammograms. In annual international conference of the IEEE
[16] Raghavendra U, Acharya UR, Fujita H, Gudigar A, Tan JH, engineering in medicine and biology society 2017 (pp. 689-692).
Chokkadi S. Application of Gabor wavelet and Locality Sensitive IEEE.
Discriminant Analysis for automated identification of breast cancer [31] Wang H, Zheng B, Yoon SW, Ko HS. A support vector machine-
using digitized mammogram images. Applied Soft Computing. based ensemble algorithm for breast cancer diagnosis. European
2016; 46:151-61. Journal of Operational Research. 2018; 267(2):687-99.
[17] Abdel-Nasser M, Melendez J, Moreno A, Omer OA, Puig D. Breast [32] Gardezi SJ, Faye I, Bornot JM, Kamel N, Hussain M. Mammogram
tumor classification in ultrasound images using texture analysis and classification using dynamic time warping. Multimedia Tools and
super-resolution methods. Engineering Applications of Artificial Applications. 2018;77(3):3941-62.
Intelligence. 2017; 59:84-92. [33] Khan S, Hussain M, Aboalsamh H, Bebis G. A comparison of
[18] Muduli D, Dash R, Majhi B. Automated breast cancer detection in different Gabor feature extraction approaches for mass classification
digital mammograms: A moth flame optimization based ELM in mammography. Multimedia Tools and Applications. 2017;
approach. Biomedical Signal Processing and Control. 2020; 76(1):33-57.
59:101912. [34] Moon WK, Chen IL, Chang JM, Shin SU, Lo CM, Chang RF. The
[19] Song R, Li T, Wang Y. Mammographic Classification Based on adaptive computer-aided diagnosis system based on tumor sizes for
XGBoost and DCNN With Multi Features. IEEE Access. 2020; the classification of breast tumors detected at screening ultrasound.
8:75011-21. Ultrasonics. 2017; 76:70-7.
[20] Muramatsu C, Hara T, Endo T, Fujita H. Breast mass classification [35] Casti P, Mencattini A, Salmeri M, Ancona A, Lorusso M, Pepe ML,
on mammograms using radial local ternary patterns. Computers in Di Natale C, Martinelli E. Towards localization of malignant sites of
Biology and Medicine. 2016; 72:43-53. asymmetry across bilateral mammograms. Computer Methods and
[21] Zhang J, Silber JI, Mazurowski MA. Modeling false positive error Programs in Biomedicine. 2017; 140:11-8.
making patterns in radiology trainees for improved mammography [36] Dong M, Lu X, Ma Y, Guo Y, Ma Y, Wang K. An efficient
education. Journal of Biomedical Informatics. 2015; 54:50-7. approach for automated mass segmentation and classification in
[22] de Lima SM, da Silva-Filho AG, Dos Santos WP. Detection and mammograms. Journal of Digital Imaging. 2015; 28(5):613-25.
classification of masses in mammographic images in a multi-kernel [37] Malik B, Klock J, Wiskin J, Lenox M. Objective breast tissue image
approach. Computer Methods and Programs in Biomedicine. 2016; classification using Quantitative Transmission ultrasound
134:11-29. tomography. Scientific Reports. 2016; 6(1):1-8.
[23] Vijayarajeswari R, Parthasarathy P, Vivekanandan S, Basha AA. [38] Peng W, Mayorga RV, Hussein EM. An automated confirmatory
Classification of mammogram for early detection of breast cancer system for analysis of mammograms. Computer Methods and
using SVM classifier and Hough transform. Measurement. 2019; Programs in Biomedicine. 2016; 125:134-44.
146:800-5. [39] Jalalian A, Mashohor S, Mahmud R, Karasfi B, Saripan MI, Ramli
[24] Jadoon MM, Zhang Q, Haq IU, Butt S, Jadoon A. Three-class AR. Computer-assisted diagnosis system for breast cancer in
mammogram classification based on descriptive CNN features. computed tomography laser mammography (CTLM). Journal of
BioMed research international. 2017;2017. Digital Imaging. 2017;30(6):796-811.
[25] Zhang X, Homma N, Goto S, Kawasumi Y, Ishibashi T, Abe M, [40] Li H, Meng X, Wang T, Tang Y, Yin Y. Breast masses in
Sugita N, Yoshizawa M. A hybrid image filtering method for mammography classification with local contour features. Biomedical
computer-aided detection of microcalcification clusters in Engineering Online. 2017 Dec;16(1):1-2.
mammograms. Journal of Medical Engineering. 2013;2013. [41] Cordeiro FR, Santos WP, Silva-Filho AG. Analysis of supervised
[26] Eltoukhy MM, Gardezi SJ, Faye I. A method to reduce curvelet and semi-supervised GrowCut applied to segmentation of masses in
coefficients for mammogram classification. In IEEE Region 10 mammography images. Computer Methods in Biomechanics and
Symposium 2014 (pp. 663-666). IEEE. Biomedical Engineering: Imaging & Visualization. 2017; 5(4):297-
[27] Wajid SK, Hussain A. Local energy-based shape histogram feature 315.
extraction technique for breast cancer diagnosis. Expert Systems [42] Sahran S, Qasem A, Omar K, Albashih D, Adam A, Abdullah SN,
with Applications. 2015; 42(20):6990-9. Abdullah A, Hussain RI, Ismail F, Abdullah N, Pauzi SH. Machine
[28] Jiang M, Zhang S, Li H, Metaxas DN. Computer-aided diagnosis of learning methods for breast cancer diagnostic. In Breast Cancer and
mammographic masses using scalable image retrieval. IEEE Surgery 2018. IntechOpen.
Transactions on Biomedical Engineering. 2014; 62(2):783-92. [43] LeCun Y, Bengio Y, Hinton G. Deep learning. Nature. 2015;
[29] Magna G, Casti P, Jayaraman SV, Salmeri M, Mencattini A, 521(7553):436-44.
Martinelli E, Di Natale C. Identification of mammography anomalies [44] Jordan MI, Mitchell TM. Machine learning: Trends, perspectives,
for breast cancer detection by an ensemble of classification models and prospects. Science. 2015; 349(6245):255-60.
based on artificial immune system. Knowledge-Based Systems.
2016; 101:60-70.

117
 International Journal of Emerging Technology and Advanced Engineering
Website: www.ijetae.com (E-ISSN 2250-2459, Scopus Indexed, ISO 9001:2008 Certified Journal, Volume 12, Issue 03, March 2022)
[45] Kooi T, Gubern-Merida A, Mordang JJ, Mann R, Pijnappel R, [58] Qiu Y, Yan S, Gundreddy RR, Wang Y, Cheng S, Liu H, Zheng B.
Schuur K, den Heeten A, Karssemeijer N. A comparison between a A new approach to develop computer-aided diagnosis scheme of
deep convolutional neural network and radiologists for classifying breast mass classification using deep learning technology. Journal of
regions of interest in mammography. In international workshop on X-ray Science and Technology. 2017; 25(5):751-63.
breast imaging 2016 (pp. 51-56). Springer, Cham. [59] Zhang X, Zhang Y, Han EY, Jacobs N, Han Q, Wang X, Liu J.
[46] Chartrand G, Cheng PM, Vorontsov E, Drozdzal M, Turcotte S, Pal Classification of whole mammogram and tomosynthesis images
CJ, Kadoury S, Tang A. Deep learning: a primer for radiologists. using deep convolutional neural networks. IEEE Transactions on
Radiographics. 2017; 37(7):2113-31. Nanobioscience. 2018;17(3):237-42.
[47] Platania R, Shams S, Yang S, Zhang J, Lee K, Park SJ. Automated [60] Jiao Z, Gao X, Wang Y, Li J. A parasitic metric learning net for
breast cancer diagnosis using deep learning and region of interest breast mass classification based on mammography. Pattern
detection (bc-droid). In proceedings of the ACM international Recognition. 2018; 75:292-301.
conference on bioinformatics, computational biology, and health [61] Chougrad H, Zouaki H, Alheyane O. Deep convolutional neural
informatics 2017 (pp. 536-543). ACM. networks for breast cancer screening. Computer Methods and
[48] Wang J, Ding H, Bidgoli FA, Zhou B, Iribarren C, Molloi S, Baldi Programs in Biomedicine. 2018; 157:19-30.
P. Detecting cardiovascular disease from mammograms with deep [62] Al-Masni MA, Al-Antari MA, Park JM, Gi G, Kim TY, Rivera P,
learning. IEEE Transactions on Medical Imaging. 2017; 36(5):1172- Valarezo E, Choi MT, Han SM, Kim TS. Simultaneous detection
81. and classification of breast masses in digital mammograms via a
[49] Pan SJ, Yang Q. A survey on transfer learning. IEEE Transactions deep learning YOLO-based CAD system. Computer Methods and
on Knowledge and Data Engineering. 2009; 22(10):1345-59. Programs in Biomedicine. 2018; 157:85-94.
[50] Sharif Razavian A, Azizpour H, Sullivan J, Carlsson S. CNN [63] Ribli D, Horváth A, Unger Z, Pollner P, Csabai I. Detecting and
features off-the-shelf: an astounding baseline for recognition. In classifying lesions in mammograms with deep learning. Scientific
proceedings of the IEEE conference on computer vision and pattern Reports. 2018; 8(1):1-7.
recognition workshops 2014 (pp. 806-813). IEEE. [64] Wang H, Feng J, Zhang Z, Su H, Cui L, He H, Liu L. Breast mass
[51] Yosinski J, Clune J, Bengio Y, Lipson H. How transferable are classification via deeply integrating the contextual information from
features in deep neural networks? arXiv preprint arXiv:1411.1792. multi-view data. Pattern Recognition. 2018; 80:42-52.
2014. [65] Yu S, Liu L, Wang Z, Dai G, Xie Y. Transferring deep neural
[52] Castrejon L, Aytar Y, Vondrick C, Pirsiavash H, Torralba A. networks for the differentiation of mammographic breast lesions.
Learning aligned cross-modal representations from weakly aligned Science China Technological Sciences. 2019; 62(3):441-7.
data. In proceedings of the IEEE conference on computer vision and [66] Cai H, Huang Q, Rong W, Song Y, Li J, Wang J, Chen J, Li L.
pattern recognition 2016 (pp. 2940-2949). Breast microcalcification diagnosis using deep convolutional neural
[53] Jiao Z, Gao X, Wang Y, Li J. A deep feature-based framework for network from digital mammograms. Computational and
breast masses classification. Neurocomputing. 2016; 197:221-31. Mathematical Methods in Medicine. 2019; 2019.
[54] Huynh BQ, Li H, Giger ML. Digital mammographic tumor [67] Kaur P, Singh G, Kaur P. Intellectual detection and validation of
classification using transfer learning from deep convolutional neural automated mammogram breast cancer images by multi-class SVM
networks. Journal of Medical Imaging. 2016; 3(3):034501. using deep learning classification. Informatics in Medicine
[55] Samala RK, Chan HP, Hadjiiski LM, Helvie MA, Cha KH, Richter Unlocked. 2019; 16:100151.
CD. Multi-task transfer learning deep convolutional neural network: [68] Albarqouni S, Baur C, Achilles F, Belagiannis V, Demirci S, Navab
application to computer-aided diagnosis of breast cancer on N. Aggnet: deep learning from crowds for mitosis detection in breast
mammograms. Physics in Medicine & Biology. 2017; 62(23):8894. cancer histology images. IEEE Transactions on Medical Imaging.
[56] Antropova N, Huynh BQ, Giger ML. A deep feature fusion 2016; 35(5):1313-21.
methodology for breast cancer diagnosis demonstrated on three [69] Spanhol FA, Oliveira LS, Petitjean C, Heutte L. A dataset for breast
imaging modality datasets. Medical Physics. 2017; 44(10):5162-71. cancer histopathological image classification. IEEE Transactions on
[57] Dhungel N, Carneiro G, Bradley AP. A deep learning approach for Biomedical Engineering. 2015; 63(7):1455-62.
the analysis of masses in mammograms with minimal user
intervention. Medical Image Analysis. 2017; 37:114-28.

118