A cost-effective computer-vision based breast cancer diagnosis

Abstract

In the last decade, there have been extensive reports of world health organization (WHO) on breast cancer. About 2.1 million women are affected every year and it is the second most leading cause of cancer death in women. Initial detection and diagnosis of cancer appreciably increase the chance of saving lives and reduce treatment costs. In this paper, we perform a survey of the techniques utilized in breast cancer detection and diagnosis in image processing, machine learning (ML), and deep learning (DL). We also proposed a novel computer-vision based cost-effective method for breast cancer detection and diagnosis. Along with the detection and diagnosis of breast cancer, our proposed method is capable of finding the exact position of the abnormality present in the breast that will help in breast-conserving surgery or partial mastectomy. The proposed method is the simplest and cost-effective approach that has produced highly accurate and useful outcomes when compared with the existing approach.

Keywords

Breast cancer computer vision mammography support vector machine (SVM)HOG features

1 Introduction

“Breast cancer is the second most leading cause of cancer death in women [1]”. “Breast cancer is a group of diseases in which cells in breast tissue change and divide uncontrolled, typically resulting in a lump or mass (Breast Cancer Facts - National Breast Cancer Foundation n.d.)”. From the last five years, the survival rate of breast cancer patients is about 60% in India compared to 90% in the United States. [3]. A long way from reality as recent advancements have significantly improved patients’ endurance contrasted with possibly 2 –3 decades ago. It is easily treated and typically has no symptoms when the tumor is small, so screening is necessary for early recognition. The painless lump is the most common physical sign. The BC increases the risk of other diseases like osteoporosis, progeria, and Werner syndrome, which causes bone loss, fracture, hair loss, and abnormality in the skin.

According to the Surveillance, Epidemiology, and End Results (SEER) summary staging system, breast cancer is divided into seven categories, as shown in Fig. 1 and there generally has four stages, namely Stages 0 &1, Stages 2 (II) & Stage 2A (IIA), Stage 3(III) A, B, &C and Stage 4 (IV) [2]. Also, there are six types of cancer indicating abnormalities present in the breast, namely CALC (calcification), CIRC (circumcised masses), SPIC (speculated masses), MISC (other ill-defined masses), ARCH (architectural distortion), and ASYM (asymmetry) as shown in Fig. 2. Accordingly, there is no prevention technique for breast cancer; however, early detection can significantly improve the result. So, it is better to screen cancer leading abnormalities mammogram images from the normal. Then, detect the position of abnormalities present in the mammogram images.

Fig. 1

Different types of breast cancer.

Fig. 2

Types of cancer indicating abnormalities and normal mammogram images (a) ARCH (b) ASYM (c) CALC (d) CIRC (e) MISC (f) SPIC (g) normal (h) normal (i) normal.

The major contribution of this research article includes the following points:

We have reviewed almost 70 papers that deal with BC detection and diagnosis.

Analysis of the previous methodology, along with its advantages and limitations, are made.

The proposed method is a novel computer-vision based and does not require high computing performance systems with expensive hardware, making the system cost-effective.

Using the HOG and SVM features, we screen the mammogram images that contain the abnormality, which indicates the BC.

Along with the detection and diagnosis of breast cancer, our proposed method can detect the exact position of the abnormality present in the breast, which coincides with the used dataset, i.e., the MIAS database. It will also help in breast-conserving surgery or partial mastectomy.

The effectiveness of the proposed methodology and model is demonstrated via figures.

The rest of the article is organized as follows: Section 2 presents the existing techniques and literature survey. The proposed methodology is discussed in Section 3. Section 4 is allocated for the results and discussions. Finally, we conclude the work by highlighting the proposed system’s achieved results and effectiveness in Section 5.

2 Literature survey

This section described and analyzed the different applications of image processing, machine learning, deep learning, clinically significant, and other biological interpretable features applied for breast cancer detection and diagnosis. Based on the clinically substantial and anatomical interpretable features, (Kumar, Srivastava, and Srivastava [4] had suggested a method for detection and classification of BC by considering the underlying tissues (epithelial, connective, nervous, and muscular) from the microscopic 1000 random microscopic biopsy images with the support of K- nearest neighboring. To improve breast cancer classification, Ting, Tan, and Sim [5] had considered a CNN. The CNNI-BCC was classified into three groups, namely healthy, malignant, and benign patients with high accuracy of 90.50%. For predicting breast cancer risk, Id et al. [6] had used a machine learning model and personal health information, namely, “Breast Cancer Risk Prediction Tool” (BCRAT). The PLCO dataset used for experimenting was collected from the National Cancer Institute.

Again image processing and machine learning were considered by Adel et al. [7] for BC diagnosis. A Support vector machine (SVM) was used to classify the combined elastogram and B- mode images. The classification accuracy for the SVM with radial basis function (RBF) kernel was about 94.12 %. Hasan et al. [8] had presented a model based on CNN for breast mass detection to minimize the overheads of manual analysis. The datasets used for the experiments were collected from different local hospitals. The performance of the proposed model was AUC-ROC (92.2%), detection accuracy (91.86%), and sensitivity (94.67%). By adopting the different local quinary patterns along with neighborhood topologies, Rampun et al. [9] had proposed a methodology for breast density classification with competitive results up to 86.13% and 82.02% accuracy for 322 and 206 mammograms collected from the MIAS database. Sadhukhan, Upadhyay, and Chakraborty [10] had defined a framework based on image processing and machine learning for the BC diagnosis with the support of SVM and KNN. The efficiency of the result obtained in the study was relatively high, i.e., 97.489%. Kadam, Jadhav, and Vijayakumar [11] had used ensemble learning, especially based on features of the stacked sparse autoencoders and softmax regression, by deploying the SSAE-SM model with promising results as 98.60% accuracy. Vaka, Soni, and K [12] had present the new technique called Deep Neural Network with Support Value (DNNS) to detect BC with high accuracy of 97.21 % along with the precision of 97.9%. Shrivastava and Bharti [1] had suggested a methodology for breast tumor detection and classification by utilizing the concept of density on different datasets, including RIDER, MIAS, and DDSM. The result was quite effective and promising, even better than the existing cutting-edge techniques, i.e., 94.4%.

In the last few year a vast number of research works have been reported on different method for detection and diagnosis of breast cancer using modern technologies such as these approaches include, image processing [2 –4], deep learning [5 –9] with shear- wave elastography [10], Deep hybrid features [11 –13], Electromagnetism-like algorithm [14], Deep belief network [15], SSAE model [16], Fuzzy c-Means (FCM) [17], Decision trees [18 –21], Random forest [19], Artificial neural network (ANN) [22 –24], Spatial diversity [25], Probabilistic neural network [26], Radial basis function neural network (RBFNN)[27], kernel-based methods, Fuzzy Gaussian mixture model (FGMM) [28], fuzzy classifiers [29], clustering algorithms [30], evolutionary computations [31] (e.g. Genetic algorithm [32 –34], simulated annealing [35]), R-CNN Classifier [36, 37], Bidirectional Recurrent Neural Networks (HA-BiRNN) [38], Transfer Learning [39 –41], Wavelet neural network [42], SVM Classifier [43 –46], AdaBoost learning [47], Semi-supervised algorithm [48], Blustering Mining [49], DNNS [50], Deep CNN [51 –59], DCNN with misclassification reduction algorithm [60].

The survey above shows that all existing methods, as tabulated in Table 1, are not cost-effective and have performance issues. Also, from the study of our best knowledge, no technique can classify and determine the exact position of the abnormality present in the breast until now. The manuscript mainly focused on the precise position detection of the various abnormality present in the breast utilizing the mammographic database [61] for the image analysis. So, breast-conserving surgery or partial mastectomy can be carried out. Further, this can impressively reduce the expenses of the treatment.

Table 1
Comparative analysis of the most popular methods for breast cancer detection and diagnosis

Application goals and scenarios Author and year Method adopted The results obtained Advantages and disadvantage

For detection and classification of BC. Kumar, Srivastava, and Srivastava 2015 [62] K- nearest neighbor (KNN) Accuracy = 92.19 %, BCR = 88.02%, MCC = 71.74 % and F-measure = 75.94 % High performance in the case of connective tissue.

For breast tumors classification. Zhang et al. 2016 [10] DL architecture with PGBM and RBM. Accuracy = 93.4%, AUC = 97.10% and sensitivity = 88.6% It can be used in clinical computer-aided diagnosis.

For thermo-graphic image classification. Oleszkiewicz et al. 2016 [44] Supervised Learning Model: SVM with SMO Average accuracy and sensitivity were obtained. Cheaper and effective.

For Nuclei detection of BC. Xu et al. 2016 [16] Stacked Sparse Autoencoder (SSAE) F- measure = 84.49% and Average Precision = 78.83% Also, perform nine other nuclear detection strategies.

For BC diagnosis. Sheikhpour, Sarram, and Sheikhpour 2016 [63] The proposed model, namely PSO-KDE Low classification error. Also solves multiple criteria issue.

To reduce the requirement for labeled data for BC diagnosis. Peng et al. 2016 [48] Artificial intelligence, i.e., semi-supervised diagnosis method. Accuracy = 98.30 %, sensitivity = 94.3%, precision = 99.6%, F-measure = 96.4 % and standard deviation = 0.11. It also has clonal selection and excellent immune characteristics.

For BC Classification. Pan et al. 2017 [37] CE + CNN+ Grid- LSTM Precision = 87.36%, Recall = 74.43% and F1 score = 80.38 %. Easy to apply various NLP tasks.

For BC Screening Chougrad, Zouaki, and Alheyane 2018 [53] Deep CNN &transfer learning. High Accuracy of 98.23 % and 0.99 AUC. Highly efficient.

For improving BC diagnosis. F. et al. 2018 [57] SD-CNN Accuracy = 95.00% Better performance than the CEDM.

To distinguish recalled benign. S.S. et al. 2018 [7] Deep learning CNN models. AUC = 96.00% Help to reduce false recalls.

For BC data. Naga RamaDevi, Usha Rani, and Lavanya 2019 [19] Synthetic Minority Over Sampling Technique(SMOTE) and Principal Component Analysis (PCA) High accuracy High performance

Breast Density Classification. Rampun et al. 2018 [64] Local Quinary pattern (LQP) and different NN. The average accuracy of 86.13% for 322 and 82.02% for 206 mammograms. Average performances.

Ultrasonic breast masses discrimination. Xiao et al. 2018 [40] Deep CNN Accuracy of 89.44% and AUC 0.93. InceptionV3 has better performances.

For improving BC Detection. Shen et al. 2019 [6] Deep learning, CBIS-DDSM, and FFDM Sensitivity = 86.7%, specificity = 96.1%. It helps in reducing the false-negative and even also in false positive.

For improving BC Detection. Ting, Tan, and Sim 2019) [52] CNNI-BCC Classification accuracy = 90.50% and specificity = 90.71%. Able to detect and classify BC.

Early prediction of BC. Id et al. 2019 [65] Machine Learning model and Breast Cancer Risk Prediction Tool (BCRAT) AUC = 0.563 and accuracy = 95% Need to train on another dataset for more improvement.

BC classification Adel et al. 2019 [66] CAD and DBN-NN Overall accuracy = 99.68%, specificity = 99.47% and sensitivity = 100% Better performances over previous methods.

BC Detection Hasan et al. 2019 [67] DCNN and SVM SVM accuracy = 87.2% and AUC = 94% Highest AUC value compared to existing methods.

BC diagnosis. Kadam, Jadhav, and Vijayakumar 2019 [63] SSAE-SM model Accuracy = 98.60% Highly efficient

BC classification Chen, Huang, and Li 2019 [38], KDBTC, BiRNN and HA-BiRNN Accuracy = 88.54%,Recall = 87.77% and Fi score = 90.70 % Better performance while discriminating BC.

Detection and Classification of BC Khan et al. 2019 [39] Deep learning like GoogleNet, VGGNet, and ResNet High classification accuracy Better performance.

To differentiate mammographic breast lesions. Yu et al. 2019 [41] Transferred DNNs and CNNs DNNs (ACC = 0.81 and AUC = 0.88)&CNN(ACC = 0.73 and AUC = 0.82) Wide application is required for verifying the medical images.

BC Diagnosis Sadhukhan, Upadhyay, and Chakraborty 2020 [68] Image Processing and Machine Learning, i.e., KNN and SVM Highly efficient about 97.489% Speedy diagnosis of BC.

BC detection Vaka, Soni, and K. 2020) [50] Machine Learning, i.e., DNNS Accuracy = 97.21 %,Precision = 97.9 and Recall = 97.01% Better performance.

Prediction of BC risk. Behravan et al. 2020 [34] Demographic factors, Genetic and Machine Learning Highly accurate. High efficiency

Breast tumor classification. Huang et al. 2020 [49] AdaBoost and Biclustering Mining Accuracy = 95.75%, sensitivity y = 96.26% and Specificity = 95.12% Hard to diagnosis some tumors

Breast Mass Classification. Fujioka et al. 2020 [55] SWE and CNN’s High accuracy Better performance when compared to SWEc (P = 0.018-0.037)

Application goals and scenarios	Author and year	Method adopted	The results obtained	Advantages and disadvantage
For detection and classification of BC.	Kumar, Srivastava, and Srivastava 2015 [62]	K- nearest neighbor (KNN)	Accuracy = 92.19 %, BCR = 88.02%, MCC = 71.74 % and F-measure = 75.94 %	High performance in the case of connective tissue.
For breast tumors classification.	Zhang et al. 2016 [10]	DL architecture with PGBM and RBM.	Accuracy = 93.4%, AUC = 97.10% and sensitivity = 88.6%	It can be used in clinical computer-aided diagnosis.
For thermo-graphic image classification.	Oleszkiewicz et al. 2016 [44]	Supervised Learning Model: SVM with SMO	Average accuracy and sensitivity were obtained.	Cheaper and effective.
For Nuclei detection of BC.	Xu et al. 2016 [16]	Stacked Sparse Autoencoder (SSAE)	F- measure = 84.49% and Average Precision = 78.83%	Also, perform nine other nuclear detection strategies.
For BC diagnosis.	Sheikhpour, Sarram, and Sheikhpour 2016 [63]	The proposed model, namely PSO-KDE	Low classification error.	Also solves multiple criteria issue.
To reduce the requirement for labeled data for BC diagnosis.	Peng et al. 2016 [48]	Artificial intelligence, i.e., semi-supervised diagnosis method.	Accuracy = 98.30 %, sensitivity = 94.3%, precision = 99.6%, F-measure = 96.4 % and standard deviation = 0.11.	It also has clonal selection and excellent immune characteristics.
For BC Classification.	Pan et al. 2017 [37]	CE + CNN+ Grid- LSTM	Precision = 87.36%, Recall = 74.43% and F1 score = 80.38 %.	Easy to apply various NLP tasks.
For BC Screening	Chougrad, Zouaki, and Alheyane 2018 [53]	Deep CNN &transfer learning.	High Accuracy of 98.23 % and 0.99 AUC.	Highly efficient.
For improving BC diagnosis.	F. et al. 2018 [57]	SD-CNN	Accuracy = 95.00%	Better performance than the CEDM.
To distinguish recalled benign.	S.S. et al. 2018 [7]	Deep learning CNN models.	AUC = 96.00%	Help to reduce false recalls.
For BC data.	Naga RamaDevi, Usha Rani, and Lavanya 2019 [19]	Synthetic Minority Over Sampling Technique(SMOTE) and Principal Component Analysis (PCA)	High accuracy	High performance
Breast Density Classification.	Rampun et al. 2018 [64]	Local Quinary pattern (LQP) and different NN.	The average accuracy of 86.13% for 322 and 82.02% for 206 mammograms.	Average performances.
Ultrasonic breast masses discrimination.	Xiao et al. 2018 [40]	Deep CNN	Accuracy of 89.44% and AUC 0.93.	InceptionV3 has better performances.
For improving BC Detection.	Shen et al. 2019 [6]	Deep learning, CBIS-DDSM, and FFDM	Sensitivity = 86.7%, specificity = 96.1%.	It helps in reducing the false-negative and even also in false positive.
For improving BC Detection.	Ting, Tan, and Sim 2019) [52]	CNNI-BCC	Classification accuracy = 90.50% and specificity = 90.71%.	Able to detect and classify BC.
Early prediction of BC.	Id et al. 2019 [65]	Machine Learning model and Breast Cancer Risk Prediction Tool (BCRAT)	AUC = 0.563 and accuracy = 95%	Need to train on another dataset for more improvement.
BC classification	Adel et al. 2019 [66]	CAD and DBN-NN	Overall accuracy = 99.68%, specificity = 99.47% and sensitivity = 100%	Better performances over previous methods.
BC Detection	Hasan et al. 2019 [67]	DCNN and SVM	SVM accuracy = 87.2% and AUC = 94%	Highest AUC value compared to existing methods.
BC diagnosis.	Kadam, Jadhav, and Vijayakumar 2019 [63]	SSAE-SM model	Accuracy = 98.60%	Highly efficient
BC classification	Chen, Huang, and Li 2019 [38],	KDBTC, BiRNN and HA-BiRNN	Accuracy = 88.54%,Recall = 87.77% and Fi score = 90.70 %	Better performance while discriminating BC.
Detection and Classification of BC	Khan et al. 2019 [39]	Deep learning like GoogleNet, VGGNet, and ResNet	High classification accuracy	Better performance.
To differentiate mammographic breast lesions.	Yu et al. 2019 [41]	Transferred DNNs and CNNs	DNNs (ACC = 0.81 and AUC = 0.88)&CNN(ACC = 0.73 and AUC = 0.82)	Wide application is required for verifying the medical images.
BC Diagnosis	Sadhukhan, Upadhyay, and Chakraborty 2020 [68]	Image Processing and Machine Learning, i.e., KNN and SVM	Highly efficient about 97.489%	Speedy diagnosis of BC.
BC detection	Vaka, Soni, and K. 2020) [50]	Machine Learning, i.e., DNNS	Accuracy = 97.21 %,Precision = 97.9 and Recall = 97.01%	Better performance.
Prediction of BC risk.	Behravan et al. 2020 [34]	Demographic factors, Genetic and Machine Learning	Highly accurate.	High efficiency
Breast tumor classification.	Huang et al. 2020 [49]	AdaBoost and Biclustering Mining	Accuracy = 95.75%, sensitivity y = 96.26% and Specificity = 95.12%	Hard to diagnosis some tumors
Breast Mass Classification.	Fujioka et al. 2020 [55]	SWE and CNN’s	High accuracy	Better performance when compared to SWEc (P = 0.018-0.037)

3 Methodology

For early diagnosis of breast cancer, it is required to screen the abnormal mammogram images and then detect the position of cancer indicating abnormalities. For this, we made two groups to all mammogram images, i.e., normal and abnormal. The abnormal includes all six cancer indicating abnormal mammogram images. The proposed approach as a block diagram is depicted in Fig. 3. The methodology comprises two phases. In the first phase, it screens the abnormal breast from the normal one. For this, 113 abnormal images and 209 normal images are used. The HOG features of both kinds of images are extracted and fed to SVM for classification. In the second phase, the abnormal breasts are considered for image analysis to detect the position of presence of abnormal tissue. The exploration study was actualized utilizing the MATLAB 2019a. All the applications were run on a laptop, i.e., HP Pavilion Core i5 5th Generation.

Fig. 3

Flow chart for the proposed approach.

3.1 Image collection

For experimentation, we have used images of the mammography database version 1.21. According to the Mammographic Image Analysis Society (J Suckling et al. 1994), there are six different cancer types: CALC, ASYM, CIRC, SPIC, MISC, and ARCH.

3.2 Breast and muscle extraction

The Breast region is cropped for extracting the clear sight of breast and muscle and labeled as “Getting the Breast and Muscle”. After that, we deleted the black corners to select the muscle by using the ‘bwselect’ function, converting it into B&W, and removed. Also, we remove the muscle part too for getting a breast image only for the analysis.

3.3 Filtering out noise

For getting the fine and clear image of the breast only for proper analysis, we filter out the noise, i.e., noise spectra and additive noise, by using a wiener filter concerning the window size and SNR of 3 x 3 and 02. Respectively. Along with filtering out the noise from the image, the Wiener filter is also utilized for the evacuation of obscure images due to linear motion or unfocussed optics.

3.4 AHE

An image contrast enhancement algorithm called Adaptive Histogram Equalization (AHE) is being utilized to overcome the limitations in the standard histogram equalization. The contrast limited AHE (CLAHE) to reduce noise by partially reducing the standard histogram equalization. Here we used the CLAHE. Also, HOG features are extracted from the CLAHE passed image.

3.5 SVM

After extracting the HOG features, SVM is deployed, which is a supervised machine learning model. It is used to classify the normal and abnormal parts, as shown in the flow chart depicted in Fig. 3.

3.6 Determine the large Blob

As the normal and abnormality classification is done, the abnormal once again goes through the process, including threshold, Adaptive histogram equalization for determining the large Blob as shown in Fig. 3.

3.7 Exact position determination

Finally, after the extraction of the large Blob, the exact position of the affected area from breast cancer is determined by deploying the blue colored bounding box. Hence, the exact location is detected of the various abnormality present in the breast.

4 Results and discussions

Here in this article, we proposed a novel computer-vision based cost-effective method for breast cancer detection and diagnosis. The Mammographic database used for the experiment consists of 322 images, and about 99.64% of accuracy achieved to screen abnormal mammogram images from the normal. The exact position of the abnormality present in the mammogram images is determined, which is useful for the BC diagnosis.

Figures 4, 5 and Fig. 6 illustrate the process of detection of the position of abnormality. The red circle is the position of abnormality provided by the radiology specialist, and the blue rectangle is the position of abnormality detected by our proposed method. For our observation, all images available in the MIAS version 1.21 database of ASYM, CIRC, and ARCH are considered. The proposed methodology successfully detects the position of abnormality in mammogram images. The adapted method in the state-of-art achieved satisfactory outcomes is based on a deep learning approach for BC diagnosis. Deep Learning requires a lot of hardware. Some train simple deep learning models for days on their laptops (typically without GPUs), which leads to an impression that Deep Learning requires big systems to execute. Again, the deep learning models are resource-intensive and time-consuming, as a huge number of matrix multiplications are necessary for the training phase. The training time is possibly reduced with the use of GPU but expensive. Another way, instead of buying an expensive GPU, we can leverage GPUs on-demand with a cloud-hosting company where they demand a minimum of US$0.25 per hour. In contrast, our proposed system was competent enough with the state-of-art without any high-cost GPU, which made the system cost-effective.

Fig. 4

Processed outputs to detect position of architectural abnormality (a) original image (b) binarize image (c) region of breast and muscle (d) background removed image (e)removed the muscles (f) background removed (g) filtered output (h) clear boarder (i) extract large blob (j) position of abnormality detected.

Fig. 5

Processed outputs to detect position of asymmetry abnormality (a) original image (b) binarize image (c) region of breast and muscle (d) back-ground removed image (e)removed the muscles (f) background removed (g) filtered output (h) clear boarder (i) extract large blob (j) position of abnormality detected.

Fig. 6

Results for circumscribed masses abnormality (a) original image (b) binarize image (c) region of breast and muscle (d) back-ground removed image (e) removed the muscles (f) background removed (g) filtered output (h) clear boarder (i) extract large blob (j) position of abnormality detected.

5 Conclusion

The proposed method is the simplest and cost-effective as the early screening, accurate prediction, and diagnosis of Breast Cancers are essential. In the early prediction of breast cancer, the computer-aided intelligent and automated medical decision support stems based on machine learning and soft computing play crucial roles. This paper presents a robust and sound classifier using the HOG feature and SVM to screen the abnormal breast mammogram images with an accuracy of 99.64%. Further, the abnormalities position is detected successfully. In addition, the proposed system does not require any training process and expensive hardware. Further, the lucrative performance of the system without any high-cost GPU made the system cost-effective.

References

Shrivastava

and Bharti

, Breast tumor detection and classification based on density, Multimedia Tools and Applications (2020), 1–21. https://doi.org/10.1007/s11042-020-09220-x

Fei

, Luo

, Liu

, Xu

, Chen

and Ji

, Analysis of the role of breast dynamic nuclear magnetic resonance imaging in the treatment of breast tumors, Multimedia Tools and Applications (2019), 1–16. https://doi.org/10.1007/s11042-019-07913-6

Guzmán-Cabrera

, Guzmán-Sepúlveda

J.R.

, Torres-Cisneros

, May-Arrioja

D.A.

, Ruiz-Pinales

, Ibarra-Manzano

O.G.

, et al., Digital image processing technique for breast cancer detection, International Journal of Thermophysics34 (2013), 1519–1531. https://doi.org/10.1007/s10765-012-1328-4

El Idrissi El Kaitouni

, Abbad

and Tairi

, A breast tumor segmentation and elimination of pectoral muscle based on hidden markov and region growing, Multimedia Tools and Applications77 (2018), 31347–31362. https://doi.org/10.1007/s11042-018-6089-z

, Pan

, Yang

, Liu

, He

, Li

, et al., Multi-task deep learning for fine-grained classification and grading in breast cancer histopathological images, Multimedia Tools and Applications79 (2020), 14509–14528. https://doi.org/10.1007/s11042-018-6970-9

Shen

, Margolies

L.R.

, Rothstein

J.H.

, Fluder

and Mcbride

, Deep Learning to Improve Breast Cancer Detection on Screening Mammography (2019), 1–12. https://doi.org/10.1038/s41598-019-48995-4

S.S. A, A.A. M,W.A. B, M.L. Z, J.H. S, S.W. Deep learning to distinguish recalled but benign mammography images in breast cancer screening, Clinical Cancer Research 2018. https://doi.org/10.1158/1078-0432.CCR-18-1115

Q. C, J. L and X. W. Deep learning mammography model from public dataset to clinical practice-performance discrepancy examined, Medical Physics 2018. https://doi.org/10.1002/mp.12938

Murtaza

, Shuib

, Mujtaba

and Raza

, Breast Cancer Multi-classification through Deep Neural Network and Hierarchical Classification Approach, Multimedia Tools and Applications79 (2020), 15481–15511. https://doi.org/10.1007/s11042-019-7525-4

10.

Zhang

, Xiao

, Dai

, Suo

, Wang

, Shi

, et al., Deep learning based classification of breast tumors with shear-wave elastography, Ultrasonics72 (2016), 150–157. https://doi.org/10.1016/j.ultras.2016.08.004

11.

, Chen

, Li

, Peng

, Li

and Yang

, Correction to: Breast cancer classification in pathological images based on hybrid features, Multimedia Tools and Applications78 (2019), 21347. https://doi.org/10.1007/s11042-019-7549-9

12.

, Chen

, Li

, Peng

, Li

and Yang

, Breast cancer classification in pathological images based on hybrid features, Multimedia Tools and Applications78 (2019), 21325–21345. https://doi.org/10.1007/s11042-019-7468-9

13.

Kriti

V.J.

and Agarwal

, Deep feature extraction and classification of breast ultrasound images, Multimedia Tools and Applications (2020), 1–36. https://doi.org/10.1007/s11042-020-09337-z

14.

Soulami

K.B.

, Saidi

M.N.

, Honnit

, Anibou

and Tamtaoui

, Detection of breast abnormalities in digital mammograms using the electromagnetism-like algorithm, Multimedia Tools and Applications78 (2019), 12835–12863. https://doi.org/10.1007/s11042-018-5934-4

15.

Abdel-Zaher

A.M.

and Eldeib

A.M.

, Breast cancer classification using deep belief networks, Expert Systems with Applications46 (2016), 139–144. https://doi.org/10.1016/j.eswa.2015.10.015

16.

, Xiang

, Liu

, Gilmore

, Wu

, Tang

, et al., Stacked sparse autoencoder (SSAE) for nuclei detection on breast cancer histopathology images, IEEE Transactions on Medical Imaging35 (2016), 119–130. https://doi.org/10.1109/TMI.2015.2458702

17.

Singh

B.K.

, Jain

, Banchhor

S.K.

and Verma

, Performance evaluation of breast lesion detection systems with expert delineations: a comparative investigation on mammographic images, Multimedia Tools and Applications78 (2019), 22421–22444. https://doi.org/10.1007/s11042-019-7570-z

18.

Bamakan

S.M.H.

and Gholami

, A novel feature selection method based on an integrated data envelopment analysis and entropy model, Elsevier B.V.31 (2014), 632–638. https://doi.org/10.1016/j.procs.2014.05.310

19.

Naga RamaDevi

, Usha Rani

and Lavanya,

, Ensemble-based hybrid approach for breast cancer data. In: Kumar A., Mozar S. (eds) ICCCE 2018. ICCCE 2018. Lecture Notes in Electrical Engineering, Springer, Singapore, vol. 500, Springer Verlag; 2019, p. 713–20. https://doi.org/10.1007/978-981-13-0212-1_72

20.

Luukka

and Leppälampi

, Similarity classifier with generalized mean applied to medical data, Computers in Biology and Medicine36 (2006), 1026–1040. https://doi.org/10.1016/j.compbiomed.2005.05.008

21.

Salama

G.I.

, Abdelhalim

M.B.

and Zeid

M.A.

, Breast Cancer Diagnosis on Three Different Datasets Using Multi- Classifiers 2012.

22.

Patel

A.R.

, Patra

, Shah

N.P.

and Shukla

, Biological control of mycotoxins by probiotic lactic acid bacteria, Dynamism in Dairy Industry and Consumer Demands2015 (2017), 2–4. https://doi.org/10.1155/2015

23.

Zhao

J.Y.

and Zhang

Z.L.

, Fuzzy rough neural network and its application to feature selection,pp, Proceedings of 4th International Workshop on Advanced Computational Intelligence, IWACI2011 (2011), 684–687. https://doi.org/10.1109/IWACI.2011.6160094

24.

Jafari-Marandi

, Davarzani

, Soltanpour Gharibdousti

and Smith

B.K.

, An optimum ANN-based breast cancer diagnosis: Bridging gaps between ANN learning and decision-making goals, Applied Soft Computing Journal72 (2018), 108–120. https://doi.org/10.1016/j.asoc.2018.07.060

25.

Braz Junior

, da Rocha

S.V.

, de Almeida

J.D.S.

, de Paiva

A.C.

, Silva

A.C.

and Gattass

, Breast cancer detection in mammography using spatial diversity, geostatistics, and concave geometry, Multimedia Tools and Applications78 (2019), 13005–13031. https://doi.org/10.1007/s11042-018-6259-z

26.

Mert

, Kılıç

and Akan

, An improved hybrid feature reduction for increased breast cancer diagnostic performance, Biomedical Engineering Letters4 (2014), 285–291. https://doi.org/10.1007/s13534-014-0148-9

27.

Kanojia

M.G.

and Abraham

, Breast cancer detection using RBF neural network, Proceedings of the 2016 2nd International Conference on Contemporary Computing and Informatics, IC3I 2016, Institute of Electrical and Electronics Engineers Inc.; (2016), pp. 363–368. https://doi.org/10.1109/IC3I.2016.7917990

28.

Aminikhanghahi

, Shin

, Wang

, Jeon

S.I.

and Son

S.H.

, A new fuzzy Gaussian mixture model (FGMM) based algorithm for mammography tumor image classification, Multimedia Tools and Applications76 (2017), 10191–205. https://doi.org/10.1007/s11042-016-3605-x

29.

Lim

C.K.

and Chan

C.S.

, A weighted inference engine based on interval-valued fuzzy relational theory, Expert Systems with Applications42 (2015), 3410–3419. https://doi.org/10.1016/j.eswa.2014.12.025

30.

Zheng

, Yoon

S.W.

and Lam

S.S.

, Breast cancer diagnosis based on feature extraction using a hybrid of K-means and support vector machine algorithms, Expert Systems with Applications41 (2014), 1476–1482. https://doi.org/10.1016/j.eswa.2013.08.044

31.

Mustra

, Grgic

and Rangayyan

R.M.

, Review of recent advances in segmentation of the breast boundary and the pectoral muscle in mammograms, Medical & Biological Engineering & Computing54 (2016), 1003–1024. https://doi.org/10.1007/s11517-015-1411-7

32.

Abed

B.M.

, Shaker

, Jalab

H.A.

, Shaker

, Mansoor

A.M.

, Alwan

A.F.

, et al., A hybrid classification algorithm approach for breast cancer diagnosis, IEACon 2016 - 2016 IEEE Industrial Electronics and Applications Conference, 2017. https://doi.org/10.1109/IEACON.2016.8067390

33.

Sheikhpour

, Sarram

M.A.

and Sheikhpour

, Particle swarm optimization for bandwidth determination and feature selection of kernel density estimation based classifiers in diagnosis of breast cancer, Applied Soft Computing Journal40 (2016), 113–131. https://doi.org/10.1016/j.asoc.2015.10.005

34.

Behravan

, Hartikainen

J.M.

, Tengström

and Kosma

V.M.

, Predicting breast cancer risk using interacting genetic and demographic factors and machine learning (2020), 1–16. https://doi.org/10.1038/s41598-020-66907-9

35.

Liu

, Qi

E.S.

, Xu

, Gao

and Liu

G.Q.

, A novel intelligent classification model for breast cancer diagnosis, Information Processing and Management56 (2019), 609–623. https://doi.org/10.1016/j.ipm.2018.10.014

36.

Bakkouri

and Afdel

, Multi-scale CNN based on region proposals for efficient breast abnormality recognition, Multimedia Tools and Applications78 (2019), 12939–12960. https://doi.org/10.1007/s11042-018-6267-z

37.

Pan

, Zhang

, Chen

and Xu

, Character-based convolutional grid neural network for breast cancer classification, Proceedings - 2017 International Conference on Green Informatics, ICGI 2017, Institute of Electrical and Electronics Engineers Inc. (2017), pp. 41–48. https://doi.org/10.1109/ICGI.2017.31

38.

Chen

, Huang

and Li

, Knowledge-Powered Deep Breast Tumor Classification with Multiple Medical Reports, IEEE/ACM Transactions on Computational Biology and Bioinformatics (2019), 1–1. https://doi.org/10.1109/tcbb.2019.2955484

39.

Khan

S.U.

, Islam

, Jan

, Ud Din

and Rodrigues

J.J.P.C.

, A novel deep learning based framework for the detection and classification of breast cancer using transfer learning, Pattern Recognition Letters125 (2019), 1–6. https://doi.org/10.1016/j.patrec.2019.03.022

40.

Xiao

, Liu

, Li

, Qin

, Yu

and Li

, Comparison of Transferred Deep Neural Networks in Ultrasonic Breast Masses Discrimination, BioMed Research International (2018), 2018. https://doi.org/10.1155/2018/4605191

41.

Yu S.

, Liu

L.L.

, Wang

Z.Y.

, Dai

G.Z.

and Xie

Y.Q.

, Transferring deep neural networks for the differentiation of mammographic breast lesions, Science China Technological Sciences 2019. https://doi.org/10.1007/s11431-017-9317-3

42.

Yahia

, Said

and Zaied

, A novel classification approach based on Extreme Learning Machine and Wavelet Neural Networks, Multimedia Tools and Applications79 (2020), 13869–13890. https://doi.org/10.1007/s11042-019-08248-y

43.

Huang

, Zhang

and Li

, Few-shot decision tree for diagnosis of ultrasound breast tumor using BI-RADS features, Multimedia Tools and Applications77 (2018), 29905–18. https://doi.org/10.1007/s11042-018-6026-1

44.

Oleszkiewicz

, Cichosz

, Jagodziński

, Matysiewicz

, Neumann

Ł.

, Nowak

R.M.

, et al., Application of SVM classifier in thermographic image classification for early detection of breast cancer, Photonics Applications in Astronomy, Communications, Industry, and High-Energy Physics Experiments (2016), 2016. https://doi.org/10.1117/12.2249063

45.

Anji Reddy

and Soni

, Breast Cancer Identification and Diagnosis Techniques. In: Rout J., Rout M., Das H. (eds) Machine Learning for Intelligent Decision Science. Algorithms for Intelligent Systems. Springer, Singapore, Springer, Singapore; 2020, p. 49–70. https://doi.org/10.1007/978-981-15-3689-2_3

46.

Gong

, Yang

, Wang

, Qi

, Guo

and Ma

, Breast density analysis based on glandular tissue segmentation and mixed feature extraction, Multimedia Tools and Applications78 (2019), 31185–31214. https://doi.org/10.1007/s11042-019-07917-2

47.

Matos

C.E.F.

, Souza

J.C.

, Diniz

J.O.B.

, Junior

G.B.

, de Paiva

A.C.

, de Almeida

J.D.S.

, et al., diagnosis of breast tissue in mammography images based local feature descriptors, Multimedia Tools and Applications78 (2019), 12961–12986. https://doi.org/10.1007/s11042-018-6390-x

48.

Peng

, Chen

, Zhou

, Li

, Yang

and Zhang

, An immune-inspired semi-supervised algorithm for breast cancer diagnosis, Computer Methods and Programs in Biomedicine134 (2016), 259–265. https://doi.org/10.1016/j.cmpb.2016.07.020

49.

Huang

, Chen

, Liu

, Tao

and Li

, On Combining Biclustering Mining and AdaBoost for Breast Tumor Classification, IEEE Transactions on Knowledge and Data Engineering32 (2020), 728–738. https://doi.org/10.1109/TKDE.2019.2891622

50.

Vaka

A.R.

, Soni

and K. SR, Breast cancer detection by leveraging Machine Learning, ICT Express 2020. https://doi.org/10.1016/j.icte.2020.04.009

51.

Saranyaraj

, Manikandan

and Maheswari

, A deep convolutional neural network for the early detection of breast carcinoma with respect to hyper- parameter tuning, Multimedia Tools and Applications79 (2020), 11013–11038. https://doi.org/10.1007/s11042-018-6560-x

52.

Ting

F.F.

, Tan

Y.J.

and Sim

K.S.

, Convolutional neural network improvement for breast cancer classification, Expert Systems with Applications 2019. https://doi.org/10.1016/j.eswa.2018.11.008

53.

Chougrad

, Zouaki

and Alheyane

, Deep Convolutional Neural Networks for breast cancer screening, Computer Methods and Programs in Biomedicine157 (2018), 19–30. https://doi.org/10.1016/j.cmpb.2018.01.011

54.

Tang

C-M.

, Cui

X-M.

, Xiang

Y.U.

and Fan

, Five Classifications of Mammography Images Based on Deep Cooperation Convolutional Neural Network, Technology, and Sciences (ASRJETS) American Scientific Research Journal for Engineering 2019.

55.

Fujioka

, Katsuta

, Kubota

, Mori

, Kikuchi

, Kato

, et al., classification of breast masses on ultrasound shear wave elastography using convolutional neural networks, Ultrasonic Imaging 2020. https://doi.org/10.1177/0161734620932609

56.

Perre

A.C.

, Alexandre

L.A.

and Freire

L.C.

, Lesion classification in mammograms using convolutional neural networks and transfer learning, Computer Methods in Biomechanics and Biomedical Engineering: Imaging and Visualization 2019. https://doi.org/10.1080/21681163.2018.1498392

57.

F. G, T. W, J. L, B. Z, L. R, D. S, et al., SD-CNN: A shallow-deep CNN for improved breast cancer diagnosis, Computerized Medical Imaging and Graphics 2018. https://doi.org/10.1016/j.compmedimag.2018.09.004

58.

Yurttakal

A.H.

, Erbay

, İkizceli

and Karaçavuş

, Detection of breast cancer via deep convolution neural networks using MRI images, Multimedia Tools and Applications79 (2020), 15555–15573. https://doi.org/10.1007/s11042-019-7479-6

59.

Sekaran

, Chandana

, Krishna

N.M.

and Kadry

, Deep learning convolutional neural network (CNN) With Gaussian mixture model for predicting pancreatic cancer, Multimedia Tools and Applications79 (2020), 10233–10247. https://doi.org/10.1007/s11042-019-7419-5

60.

Murtaza

, Shuib

, Wahab

AWA

, Mujtaba

and Raza

, Ensembled deep convolution neural network-based breast cancer classification with misclassification reduction algorithms, Multimedia Tools and Applications79 (2020), 18447–18479. https://doi.org/10.1007/s11042-020-08692-1

61.

Parker

, Ricketts

, Savage

and Stamatakis

, MAMMOGRAPHIC IMAGE ANALYSIS SOCIETY Mammographic Database 2015.

62.

Kumar

, Srivastava

and Srivastava

, Detection and Classification of Cancer from Microscopic Biopsy Images Using Clinically Significant and Biologically Interpretable Features, Journal of Medical Engineering2015 (2015), 1–14. https://doi.org/10.1155/2015/457906

63.

Kadam

V.J.

, Jadhav

S.M.

and Vijayakumar

, Breast Cancer Diagnosis Using Feature Ensemble Learning Based on Stacked Sparse Autoencoders and Softmax Regression, Journal of Medical Systems (2019), 43. https://doi.org/10.1007/s10916-019-1397-z

64.

Rampun

, Scotney

B.W.

, Morrow

P.J.

, Wang

and Winder

, Breast density classification using local quinary patterns with various neighbourhood topologies, Journal of Imaging4 (2018), 1–23. https://doi.org/10.3390/jimaging4010014

65.

Id GFS, Id GRH, Id BJN and Deng

, Predicting breast cancer risk using personal health data and machine learning models (2019), 1–17. https://doi.org/10.1371/journal.pone.0226765

66.

Adel

, Kotb

, Farag

, Darweesh

M.S.

and Mostafa

, Breast Cancer Diagnosis Using Image Processing and Machine Learning for Elastography Images, 2019 8th International Conference on Modern Circuits and Systems Technologies (MOCAST), IEEE (2019), pp. 1–4. https://doi.org/10.1109/MOCAST.2019.8741846

67.

Hasan

, Das

Barman S.

, Islam

and Reza

A.W.

, Breast cancer detection using convolutional neural networks, ACM International Conference Proceeding Series027 (2019), 254–158. https://doi.org/10.1145/3330482.3330525

68.

Sadhukhan

, Upadhyay

and Chakraborty

, Breast Cancer Diagnosis Using Image Processing and Machine Learning, Advances in Intelligent Systems and Computing vol. 937, Springer Verlag; (2020), pp. 113–27. https://doi.org/10.1007/978-981-13-7403-6_12