Sage Journals: Discover world-class research

Abstract

Breast cancer (BC) is the most prominent form of cancer among females all over the world. The current methods of BC detection include X-ray mammography, ultrasound, computed tomography, magnetic resonance imaging, positron emission tomography and breast thermographic techniques. More recently, machine learning (ML) tools have been increasingly employed in diagnostic medicine for its high efficiency in detection and intervention. The subsequent imaging features and mathematical analyses can then be used to generate ML models, which stratify, differentiate and detect benign and malignant breast lesions. Given its marked advantages, radiomics is a frequently used tool in recent research and clinics. Artificial neural networks and deep learning (DL) are novel forms of ML that evaluate data using computer simulation of the human brain. DL directly processes unstructured information, such as images, sounds and language, and performs precise clinical image stratification, medical record analyses and tumour diagnosis. Herein, this review thoroughly summarizes prior investigations on the application of medical images for the detection and intervention of BC using radiomics, namely DL and ML. The aim was to provide guidance to scientists regarding the use of artificial intelligence and ML in research and the clinic.

Keywords

Breast cancer machine learning radiomics artificial intelligence deep learning

Introduction

Breast cancer (BC) is the most prominent malignant tumour among females worldwide and it accounts of almost 10.4% of all cancers.¹ It is characterized by an aberrant, disorderly, and invasive proliferation of breast cells. Moreover, BC cells can readily escape to the circulatory or lymphatic system, where they generate new tumours and invade distant vital organs.² At present, BC is the largest contributor to cancer-related mortality among women between 20–50 years of age. Based on information from the American Cancer Society, in 2019 alone, there were an estimated 268 600 new incidences and 41 740 cancer-associated deaths in the United States.² Alarmingly, in 2020, the cancer-related mortality among women increased to 684 996 global deaths, making it the major contributor of deaths across the globe.¹ In addition, in 2021, the World Health Organization announced BC as the most prevalent form of global cancer, far exceeding lung cancer.¹ At present, these numbers are rapidly increasing, and a 50% rise is predicted over the next two decades owing to enhanced life expectancy, unhealthy diets, inadequate physical activity and detrimental substance intake (for example, alcohol).¹ Given these critical factors, there is an urgent need for extensive research in all areas of BC, from prevention to early detection and efficacious intervention.^2,3

Personalized BC interventions are highly dependent on accurate diagnosis. BC typically features discrete histological, molecular and clinical phenotypes; and it sometimes manifests with radiological heterogeneity. Nevertheless, classical detection techniques do not provide adequate information for proper BC diagnosis.⁴ Radiomics is a relatively new tool for the extensive analysis of medical images. The subsequent imaging features and mathematical analyses can then be used to generate machine learning (ML) models, which stratify, differentiate, and detect benign (b-BT) and malignant breast lesions (m-BT), which, in turn, can be used to establish risk and formulate the optimal intervention for patients with b-BT and/or m-BT. Relative to classical influential science and skilled radiologists, radiomics features offer a more precise diagnostic method of BT detection. Hence, radiomics-based explorations/research are well underway in different parts of the world.⁵

More recently, scientists increased the application of ML tools, namely, image segmentation, stratification and estimation for BC assessment. Image segmentation both identifies and extracts tumour and its adjoining normal tissue. ML tools, particularly, convolutional neural networks (CNNs), are widely employed for automated BC segmentation. For example, a previous study established a CNN-based algorithm, with a mean dice similarity coefficient of 0.85, to segment BC tumours in ultrasound images.⁶ Stratification delineates between m-BT and b-BT and predicts the probability of recurrence or metastasis. ML tools, namely, support vector machines (SVMs) and random forests (RFs), are frequently applied in BC stratification. For example, previous research established an RF-derived algorithm for breast tumour stratification of mammograph images, using a sensitivity (SEN) and specificity (SPC) of 87.5% and 91.7%, respectively.⁷ Prediction estimates the probability of tumour recurrence or metastasis, as well as treatment response. ML techniques, namely, deep neural networks (DNNs) and recurrent neural networks (RNNs), are also widely employed for BC estimation. For example, a study reported a DNN-derived method for estimating BC recurrence risk, with an area under the receiver operating characteristic curve (AUC) of 0.87.⁸

Medical imaging, namely, X-ray mammography (MG), magnetic resonance imaging (MRI) and positron emission tomography (PET), offers a non-invasive approach of extracting data on BC morphology, metabolism and blood flow. Radiomics, a progressive tool for the isolation of quantitative profile from medical images, has garnered much attention owing to its ability to retrieve both spatial heterogeneity and functional profiles of tumours. BC imaging has made remarkable recent progress, with marked advancements in both technology and imaging analyses. The common approaches of BC detection encompass MG, ultrasound (US), MRI and molecular imaging. Using these techniques, it is now possible to detect and diagnose early-stage BC, which ultimately enhanced patient prognoses. Moreover, ML and deep learning (DL) algorithms are also applied in BC image analysis, with promising results. Herein, this review presents a detailed summary of the current ML applications for BC imaging, namely, MG, ultrasound, MRI, PET and others. The article selection process is detailed in supplemental information. Based on the module and search string in PubMed® (see supplementary materials, Table S1) and Web of Science™ (see supplementary materials, Table S2), a retrieval strategy of the above respective databases was established (see supplementary materials, Table S1). Overall, 158 articles were selected from the date of inception of the databases up to 31 May 2023. Figure 1 shows the results of the literature search and selection.

Figure 1.

Flow diagram showing the identification of the research articles describing the current machine learning applications for breast cancer imaging that were included in this review.

X-ray mammography

X-ray mammography is routinely used to screen and diagnose BC, and it is a well-established method of minimizing cancer-associated deaths.⁹ BC tumours are initially screened using X-ray mammography, prior to manual interpretation by a radiologist to estimate whether it is b-BT and m-BT. The popularity of MG stems from its reduced-dose X-ray, enhanced-contrast, augmented-resolution detectors and X-ray system particularly intended for breast imaging. In general, MG can be separated into two groups: film screen MG (FSM), whereby a film screen serves as the end-recording device; and full-field digital MG (FFDM), whereby digital detectors serve as the recording media. Digital images from FFDM provide numerous benefits over FSM, particularly in terms of ease of image processing and enhancement.⁴

In recent times, there has been a considerable rise in artificial intelligence (AI) algorithm use in medicine. The available AI includes radiomics and DL, which markedly enhances lesion detection and diagnosis from medical images in clinical practice.^10,11 The current AI is rather sophisticated and approaches the performance of a skilled radiologist, particularly in MG.¹² MG images, photographed and quantified using radiomics, offer a remarkable diagnostic capacity for b-BT and m-BT, and they also provide other relevant data to radiologists.⁴ To further augment MG screening accuracy, the computer-aided diagnosis system (CAD) software 4 was designed for BC tumour identification and used since the 1990s. Regrettably, information from earlier CAD systems did not yield any marked improvement in patient outcome.^13,14 With significant advancements in DL visual object identification, and multiple other domains, more scientists are showing interest in establishing DL tools to augment the MG screening of women.^15–18 Emerging evidence suggested that the performance of a DL-based CAD system was comparable to the radiologist performance alone, and enhanced the radiologists’ performance in a supporting mode.^19,20 One study introduced a collaborative DL method that clearly classified pathological images into cancerous and non-cancerous BC tissues on 544 complete whole slide images, with SEN and accuracy of 97.73% and 95.29%, respectively.¹⁹

Machine learning is a form of AI introduced in the 1980s. ML primarily examines how computers mimic human learning behaviours, procure novel information, enhance existing information, as well as their own performance. ML conducts tasks without clear programming directions, namely, it identifies hidden associations between data and analyses them. Among the common ML applications are logistic regression (LR), linear regression, decision trees, RFs, naive Bayes and K-means cluster analyses, multilayer perceptron (MLP) and SVMs.²¹ Artificial neural networks (ANN) and DL are novel areas of ML and these tools analyse data using computer simulation of the human brain. ANN is based out of the biological learning mode of human brain neurons that interconnect with one another, and regulate cascade, alteration and classification. DL is more advanced than ANN. DN utilizes the hierarchical ANN to establish highly complicated learning models that elucidate data in various dimensions. DL encompasses a dynamic Bayesian network, CNN and RNN. The CNN algorithm distinctly benefits image processing and is employed in feature isolation and analysis of clinical imaging information. Recursive neural network algorithms dynamically monitor disease via time series data analysis. Traditional ML requires feature extraction from the original data, as well as processing into structured datasets, which cannot directly process unstructured data. DL also directly processes unstructured information, such as images, sounds and language; and is highly beneficial in clinical image stratification, medical record analysis and tumour diagnosis.^22,23 Thus, multiple strategies are implemented in mammogram diagnosing. These can be classified into statistical-, wavelets-, Markovian- and machine-learning-based methods.²⁴ A previous study developed a region of interest (ROI)-based CNN termed You Only Look Once, which simultaneously identified and categorized breast masses into digital mammograms.²⁵ This model was composed of distinct phases, namely, preprocessing, feature retrieval, mass detection with belief and tumour stratification with fully connected neural networks. However, this model was not appropriate for a small dataset and the tumour area was not partitioned within mammograms. To classify MG mass lesions and examine the CNN model, researchers developed deep convolutional neural networks.²¹ Upon preprocessing and normalization of isolated ROIs from the entire mammogram, the ROIs were integrated to form a unified dataset, which was then used to update the CNN. However, this technique did not minimize noise or muscles, which potentially resulted in mis-stratification.

Machine leaning, particularly breast imaging CAD system, aids in BC tumour diagnosis, and it is not influenced by the radiologist's reading mode, fatigue, distraction and other factors. Therefore, this system greatly enhances SEN of BC diagnosis.²² Researchers examined an AI system’s aptitude to substitute for doctors in BC diagnosis.¹⁹ The study revealed that the AI systems are quite efficient, relative to radiologists, in BC detection.¹⁹ Another study also evaluated cytological image analysis for BC identification and stratification using Naive Bayesian and ANN, demonstrating 98% precision in BC detection.²³ Similarly, Another study reported an enhanced crow search optimized extreme learning machine process, with approximately 98.26%, 97.193% and 98.137% accuracies for the reast, DDSM and MI - AS datasets, respectively.²⁶ Researchers developed a CAD system involving two components as follows: the first component identified an ROI; and the second extracted relevant profiles utilizing CNN.²⁷ Finally, using support vector machine, 87.2% accuracy was achieved in BC estimation from mammograms.²⁷ Another study also presented a CAD system for BC detection from MG images.²⁴ Their method manipulated MG images for ROI identification, relevant profile extraction and optimization, and BC stratification using a specific ML algorithm, SVM.²⁴ Based on their results, the aforementioned approach exhibited strong efficacy, particularly in distinguishing between normal and abnormal tumours with 100% accuracy.²⁴ Lastly, researchers employed an integrated regression learning, SVM and MLP system to stratify mammograms.²⁸ They yielded 99.42% accuracy with the Wisconsin Breast Cancer Dataset.²⁸ This current review extracted and organized the data in tabular form and summarized the use of MG in the diagnosis of breast cancer (Table 1).^{2,4,5,7,9,24,29–79}

Table 1.

Models, classes and performance for breast X-ray mammography data in selected papers.^{2,4,5,7,9,24,29–79}

Paper reference	Models/algorithm	Binary or multiclass	Classes	Accuracy	Other performance evaluation parameters	Anomaly application/task
²⁹	LIBSVM	Binary	Positive/negative	–	AUC = 0.725 ± 0.018	BC risk detection
³⁰	SVM	Binary	Malignant/benign	99.385%	Sensitivity = 100%Specificity = 98.785%	BC detection
³¹	SVM, ELM	Multiclass	Normal/benign/malignant	94.11%	–	BC detection
³²	DL, SAE	Binary	Malignant/benign	89.7%	AUC = 0.90	BC detection
³³	dANN	Binary	Malignant/benign	0.81	AUC = 0.85	BC detection
³⁴	DCNN	Binary	Malignant/benign	0.85	AUC = 0.84	BC detection
³⁵	DNN	Binary	Malignant/benign	–	–	BC tumour detection
³⁶	SVM, LASSO, KNN, B:SWIMS	Binary	Malignant/benign	0.738	Sensitivity = 0.957	BC tumour detection
³⁷	SVM	Binary	Malignant/benign	0.90	ROC = 0.95	BC tumour detection
³⁸	SVM, KNN, LDA	Binary	Malignant/benign	0.80	Specificity = 0.94 ± 0.06Sensitivity = 0.66 ± 0.24	BC diagnosis
⁷	RF	Binary	Malignant/benign	0.93	Sensitivity = 0.91Accuracy = 0.82	BC tumour detection
⁴	SVM, KNN, LR	Binary	Malignant/benign	0.978, 0.975	Specificity = 0.975Sensitivity = 0.983	BC diagnosis
³⁹	RF	Binary	Normal/tumour	0.61	Sensitivity = 70%Specificity = 49%	BC risk outcome
⁴⁰	CNN, VGG16, residual network	Binary	Normal/cancer	0.98	Specificity = 0.91Sensitivity = 0.87	BC tumour detection
⁴¹	Faster R-CNN	Binary	Lesion/non-lesionMalignant/benign	–	–	BC tumour detection
⁴²	SVM, VGG16, CNN	Binary	Malignant/benign	0.88	AUC = 0.88	BC image classification
⁵	SVM, RF, NB	Binary	Malignant/benign	0.72	AUC = 0.79	BC image classification
⁴³	DCNN	Binary	Malignant/benign	–	–	BC image classificationBC diagnosis
⁴⁴	Faster R-CNN30, CNN	Binary	Malignant/benign	0.94	Specificity 0.92	Cancer subtype classification
⁴⁵	DNN	Binary	Malignant/benign	0.92	AUC = 0.92	BC image classification
⁴⁶	ANN	Binary	Malignant/benign	0.82	AUC = 0.83	Identify risk categories
⁴⁷	VGG16, SVM, KNN, NB, DT	Binary	Malignant/benign	0.99	Specificity = 0.98	BC tumour detection
⁴⁸	SVM	Binary	Malignant/benign	0.82	AUC = 0.80	BC tumour detection
²	DCNN, VGG19	Binary	Malignant/benign	0.99	Specificity = 1.00	BC image classification
⁴⁹	DCNN	Binary	Malignant/benign	0.80	ROC 0.84	BC tumour detection
⁵⁰	RF	Multiclass	Normal/benign/malignant	0.85	F-score 0.98	BC tumour detection and classification
⁵¹	SVM, QGA	–	–	–	–	BC tumour edge detection
⁵²	DNN, AlexNet, VGGNet	Binary	Malignant/benign	0.987	Specificity = 0.99Sensitivity = 0.99	BC classification
⁵³	RBFNN	Binary	Malignant/benign	0.98	Sensitivity = 0.98	Breast microcalcifications early diagnosing
⁵⁴	DCNN	Binary	Malignant/benign	–	AUC = 0.91	Breast mass classification
⁵⁵	ANN	Binary	Malignant/benign	–	Specificity = 0.94	BC screening and diagnosis
⁵⁶	ANN, SVM	Binary	Malignant/benign	0.99	Specificity = 0.99Sensitivity = 0.99	BC tumour detection
²⁴	SRG, SVM	Binary	Malignant/benign	0.87	Specificity = 0.86Sensitivity = 0.90	BC tumour detection
⁵⁷	SVM, NSEA	Multiclass	BI-RADS classification	0.94	Time = 0.68 s	BC diagnosis
⁵⁸	MOD-RES	Binary	Malignant/benign	0.89	F-score = 0.89	BC detection
⁵⁹	DCNN	Binary	Malignant/benign	0.99	F-score = 0.99	BC detection
⁶⁰	MK-SVM	Multiclass	Normal/malignant/benign	0.96	Precision = 0.94	BC image classification
⁶¹	Kernel-SVM	Binary	Malignant/benign	0.99	–	BC detection
⁶²	SVM, ResNet50	Binary	Malignant/benign	0.74	AUC = 0.82	BC detection
⁶³	MobileNet	Binary	Malignant/benign	0.86	Sensitivity = 0.93	BC detection
⁶⁴	SCNN, DCNN	Binary	Malignant/benign	0.951	F-score = 0.957	BC diagnosis
⁶⁵	IMPA-ResNet50	Binary	Malignant/benign	0.98	AUC = 0.9788	BC diagnosis
⁶⁶	ResNet50	Binary	Malignant/benign	–	AUC = 0.866 ± 0.015	Breast lesion classification
⁶⁷	SVM, KNN, DT, LDA	Binary	Malignant/benign	0.963	Sensitivity = 94.1%Specificity = 98.2%	BC detection
⁶⁸	SqueezeNet, FSVM	Binary	Malignant/benign	0.98	F-score = 97.87%	BC detection and classification
⁶⁹	EDNN, MSVM	Multiclass	Normal/benign/malignant	96.72%	Recall = 96.24%	BC detection
⁷⁰	SVM, DT, RF, NB	Binary	Malignant/benign	0.99	–	BC detection
⁷¹	2D V-net 64 CNN	Binary	Malignant/benign	0.98	Sensitivity = 0.96Specificity = 0.96	BC detection
⁷²	AlexNet, VGG, GoogleNet CNNs	Binary	Malignant/benign	98.80%	Sensitivity = 99.62%	Breast lesion diagnosis
⁷³	LASSO regression	Binary	Malignant/benign	–	AUC = 0.940	BC diagnosis
⁷⁴	ANN	Binary	Malignant/benign	100%	Specificity = 100% Sensitivity = 100%	BC classification
⁷⁵	Classification ML	Binary	Cancer or non-cancer	–	Sensitivity = 100%	BC diagnosis
⁷⁶	LR	Binary	Negative and positive	–	AUC = 0.742Sensitivity = 0.783Specificity = 0.630	BC ALN detection
⁷⁷	VGG16, SVM	Binary	Malignant/benign	–	AUC = 0.876 ± 0.031	BC detection
⁹	Mask RCNN	Binary	Malignant/benign	–	AUC = 0.89Sensitivity = 0.93	Breast AD diagnosis
⁷⁸	QNN, CCNN	Binary	Malignant/benign	–	–	BC detection
⁷⁹	DT, ANN	Binary	Malignant/benign	93%	Sensitivity = 95%	Breast microcalcification diagnosis

LIBSVM, a library for support vector machines; AUC, area under the curve; BC, breast cancer; SVM, support vector machine; ELM, extreme learning machines; DL, deep learning; SAE, stacked autoencoder; dANN, deep learning with artificial neural network; DCNN, deep convolutional neural network; DNN, deep neural network; LASSO, least absolute shrinkage and selection operator; KNN, k-nearest neighbour; B:SWIMS, bootstrapped stagewise model selection; ROC, receiver operating characteristic; LDA, linear discriminant analysis; RF, random forests; LR, logistic regression; CNN, convolutional neural network; VGG16, visual geometry group 16; R-CNN, region-based convolutional neural network; NB, naïve Bayes; ANN, artificial neural network; DT, decision tree; VGG19, visual geometry group 19; QGA, quantum genetic algorithm; AlexNet, Alex network; VGGNet, visual geometry group network; RBFNN, radial basis function neural network; SRG, seeded region growing; NSEA, novel spectral extraction algorithm; MOD-RES, modified version of residual network 50; MK-SVM, multi-kernel support vector machine; ResNet 50, residual network 50; MobileNet, mobile network; SCNN, shallow convolutional neural network; IMPA, improved marine predators algorithm; SqueezeNet, squeeze network; FSVM, fuzzy support vector machine; EDNN, ensemble deep neural network; MSVM, multiclass support vector machine; CNNs, convolutional neural networks; ML, machine learning; ALN, axillary lymph node; AD, architectural distortion; QNN, quantum neural network; CCNN, classical convolutional neural network.

Ultrasonography

Ultrasonography is most frequently imaging technique used to screen for early-stage BC. US images are typically preferred as they, unlike MG, are not associated with radiation or compression. Nevertheless, its SEN is similar to digital MG (DM) and multiple studies confirmed its enhanced identification of invasive and node-negative BC.^80,81 Breast US (BUS) images are known to reliably delineate between b-BT and m-BT in five distinct areas, namely, shape, orientation, margin, echo patterns and posterior acoustic profiles.⁸² Despite the aforementioned advantages, US images can have some major shortcomings, namely, reduced resolution, contrast and blurred edges owing to noise originating from speckle and acoustic shadowing. Therefore, elucidation and BC identification from BUS is often challenging and operator dependent. CAD systems are an alternative automated approach for BC classification and it minimizes dependency on the operator.⁸³ BUS SEN and SPC can potentially be further augmented using ML. ML serves as a second reader that assists radiologists in forming a diagnosis.^84,85 ML algorithms study profiles of known cases to generate models that accurately diagnose patients with undetermined diseases. LR and naïve Bayes are two such programs with distinct learning capacities, and they show similar BC diagnosis on sonographic images.⁸⁶ These programs were integrated to generate an augmented model, which was grayscale profile-trained only with serial adaptive boosting.⁸⁴ Rather than boosting homogenous weak learner like canonical AdaBoost, the integrated program boosts heterogeneous classifiers like naïve Bayes and LR which have established performances, and which closely agree on individual cases.^84–86

Given that the BUS and elastography ultrasound (EUS) integration automatically provides more data for BC diagnosis, the bimodal US-based CAD has generated much attention in recent times. A previous study reported the application of a deep polynomial network on dual modal BUS and EUS properties to delineate between m-BT and b-BT tumours.⁸⁷ Another study also employed pretrained bi-channel CNNs to study and integrate bimodal US image profiles for BC tumour detection.⁸⁸ Although the aforementioned studies indicate a strong efficacy of bimodal tools, the single-mode BUS-derived CAD offers more flexibility and broader application, and its performance can be further enhanced with transfer learning using EUS as the source domain.

Artificial intelligence, composed of ML and brain-inspired DL neural networks, have great potential in combating current challenges within the global healthcare system. Recent AI applications in medical US images include numerous specific tasks, such as image segmentation and biometric measurement.⁸ Prior investigations successfully established tools for automated BT segmentation as well as CAD US intranodal vascularity quantification.⁸⁹ AI complements radiomics, which is used to extract meaningful imaging information, such as textural and wavelet data, which cannot otherwise be acquired by humans. Using the aforementioned radiomics profiles, one can train AI to conduct its own diagnosis, for example, stratifying a tumour as b-BT or m-BT. The precise identification of b-BT or m-BT from US imaging is critical for the timely intervention of BC patients. Moreover, effective decision support systems can markedly enhance radiological diagnostic ability.

A retrospective investigation evaluated US imaging and clinical information from 140 surgically confirmed BC cases.⁹⁰ In particular, the study examined twelve US and colour Doppler images using ML tools, among which eight profiles were statistically different between the two images (P < 0.05).⁹⁰ The final diagnostic exhibited an AUC of 0.88, SEN of 86.96% and SPC of 82.91%.⁹⁰ Another study established an automated BC diagnostic system with enhanced precision that could be readily run on smartphones.⁹¹ Upon US report entry, the program automatically analyses and detects lesions from all uploaded images.⁹¹ The aforementioned process consists of three subsystems.⁹¹ Firstly, noise is minimized within the images, followed by the reconstruction of high-quality images.⁹¹ Subsequently, the initial subsystem is generated based on a stacked denoising autoencoder framework and generative adversarial network.⁹¹ Then, the image is stratified according to malignancy or non-malignancy.⁹¹ DCCN is used to isolate significant profiles from the image.⁹¹ Lastly, system performance anomalies are examined and eliminated, which further diminishes the false negative rate.⁹¹ The current review extracted and organized the data in tabular form and summarized the application of US images in breast cancer diagnosis (Table 2).^{6,8,80,83,90,92–118}

Table 2.

Models, classes and performance for breast ultrasound image data in selected papers.^{6,8,80,83,90,92–118}

Paper reference	Models/algorithm	Binary or multiclass	Classes	Accuracy	Other performance evaluation parameters	Anomaly application/task
⁹²	ML, SL	Binary	Malignant/benign	0.957	Sensitivity = 0.912Specificity = 0.966	BC diagnosis
⁸³	BPANN, SVM	Binary	Malignant/benign	98.966% ± 0.527	AUC = 98.976% ± 0.473	Breast tumour classification
⁸⁰	ML	Binary	Malignant/benign	–	AUC = 0.86	BC diagnosis
⁹³	DFCN, CNN, SVM	Binary	Malignant/benign	71.9%	AUC = 0.795	BC detection
⁹⁴	SVM	Multiclass	Normal/benign/malignant	94.4%	AUC = 0.98	BC detection and classification
⁹⁰	ML	Binary	TN subtype/nTN subtype	–	AUC = 0.88Sensitivity = 86.96%Specificity = 82.91%	BC subtypes
⁹⁵	FK, FS, FB, KS, KB, SB	Binary	Malignant/benign	–	κ ≈ 0.944	BC detection
⁸⁶	ML	Binary	Malignant/benign	–	ROC = 0.958 ± 0.013	BC detection
⁹⁶	CNN	Binary	Malignant/benign		AUC = 0.9468Sensitivity = 0.88Specificity = 0.876	BC detection
⁹⁷	HSIC, SVM, DSPTC	Binary	Malignant/benign	91.44 ± 1.51%	Sensitivity = 90.29 ± 3.17%Specificity = 92.65 ± 5.06%AUC = 0.946	BC detection
⁹⁸	LR, RF, SV, XGBoost	Binary	Malignant/benign	82.1%	AUC = 0.906	BC diagnosis
⁹⁹	ML	–				BC diagnosis
¹⁰⁰	RF, LR, SVM, AdaBoost, DT	Binary	Malignant/benign	0.974	AUC = 0.97F1-score = 0.94	Breast tumour classification
⁸	RF, CNN	Binary	Malignant/benign	86%	Sensitivity = 84%Specificity = 88%F1-score = 0.83	Breast lesion classification
⁶	DCNN	Binary	Malignant/benign	86.5%	AUC = 0.875Sensitivity = 86.6%Specificity = 87.1%	BC classification
¹⁰¹	SVM	Binary	Malignant/benign	0.939 ± 0.039	Sensitivity = 0.922 ± 0.045AUC = 0.984 ± 0.012	BC detection
¹⁰²	SVM		Malignant/benign	86.0%	AUC = 0.885	Breast lesion classification
¹⁰³	DCNN	Binary	Malignant/benign	93.53%	Sensitivity = 94.42 %Specificity= 90.75 %	Breast lesion classification
¹⁰⁴	XGBoost, RF, SVM	Binary	Malignant/benign	88.4%	Sensitivity = 90.3%Specificity = 86.7%AUC = 0.890	BC diagnosis
¹⁰⁵	DCNN	Multiclass	Subclasses	–	–	BC subtype identification
¹⁰⁶	NB, SVM	Binary	Malignant/benign	89.17%	AUC = 0.95	Breast tumour classification
¹⁰⁷	KNN	Binary	Malignant/benign	70.00%	Sensitivity = 70.00%Specificity = 70.83%	BC detection and classification
¹⁰⁸	CNN	Binary	Malignant/benign	0.8186	Sensitivity = 0.8095Specificity = 0.8265	BC diagnosis
¹⁰⁹	LR, DT, RF, Bagging	Binary	DCIS, IDC	–	AUC = 0.66–0.78	BC diagnosis
¹¹⁰	NN, LR	Binary	Malignant/benign		AUC = 0.93Sensitivity = 100.0%	BC diagnosis
¹¹¹	PNN	Binary	Negative/positive	89.72%	Sensitivity = 84.00%Specificity = 94.74%	BC LN detection
¹¹²	CKHA, VGG-16, VGG-19, SqueezeNet	Multiclass	Malignant/benign/normal	97.09%	Sensitivity = 95.54%Specificity = 97.65%	BC Diagnosis and classification
¹¹³	LR, SVM	Multiclass	BI-RADS category 4A/4B/4C	–	AUC = 0.908	BI-RADS 4 lesion screening
¹¹⁴	DNN, RF	Binary	Malignant/benign	78.5%	–	BC diagnosis
¹¹⁵	DNN, SVM	Binary	Malignant/benign	87.1% ± 3.3%	Sensitivity = 77.4% ± 11.8%Specificity = 92.4% ± 7.2%	BC diagnosis
¹¹⁶	SVM, XGBoost, RF, LR, NB, KNN, MLP, CNN, LSTM, LDA	Binary	Negative/positive	0.891	AUC = 0.952Kappa = 0.763	BC SLN detection
¹¹⁷	XGBoost, CNN, ResNet	Binary	Malignant/benign	–	AUC = 0.84	BC detection
¹¹⁸	DL, ML	Binary	Malignant/benign	–	–	BC detection

ML, machine learning; SL, supervised learning; BC, breast cancer; BPANN, back-propagation artificial neural network; SVM, support vector machine; AUC, area under the curve; DFCN, dilated fully convolutional network; CNN, convolutional neural networks; TN, triple-negative; nTN, non-triple-negative; FK, forest-k-nearest neighbour; FS, forest-stochastic gradient; FB, forest-naïve Bayes; KS, k-nearest neighbour-stochastic gradient; KB, k-nearest neighbour-naïve Bayes; SB, stochastic gradient-naïve Bayes; ROC, receiver operating characteristic; HSIC, Hilbert-Schmidt Independence Criterion; DSPTC, doubly supervised parameter transfer classifier; LR, logistic regression; RF, random forests; SV, support vector; XG Boost, extreme gradient boosting; Adaboost, adaptive boosting; DT, decision tree; DCNN, deep convolutional neural networks; NB, naïve Bayes; KNN, k-nearest neighbour; DCIS, ductal carcinoma in situ; IDC, invasive ductal carcinoma; NN, neural network; PNN, probabilistic neural network; LN, lymph node; CKHA, chaotic krill herd algorithm; VGG-16, visual geometry group-16; VGG-19, visual geometry group-19; SqueezeNet, squeeze network; BI-RADS, breast imaging-reporting and data system; DNN, deep neural network; MLP, multilayer perceptron; LSTM, long short-term memory; LDA, linear discriminant analysis; SLN, sentinel lymph node; ResNet, residual network; DL, deep learning.

MRI

Breast MRI is frequently employed for the presurgical assessment of malignancy status, as well as the identification of potential ipsi- or contralateral BT. MRI-identified surplus BTs are BTs detected on presurgical MRI, which were undetected in prior MG or US. A secondary or targeted US is typically conducted to further assess the surplus BT. But, US is generally nonspecific for malignant tissue identification, and the MRI and US association rates are often widely variable between 23%–89%, based on multiple factors, including, radiologist performance or patient-specific differences.¹¹⁹

Researchers utilized deep transfer learning CAD-based diagnosis to identify BC.¹²⁰ In particular, they used multi-parametric MRI (mpMRI) relative to a simple independent CAD.¹²⁰ They demonstrated that the mpMRI tool provided superior results in delineating b-BT from m-BT; and the dynamic contrast-enhanced (DCE) image precision was 85%.¹²⁰ Nevertheless, there are certain limitations to this method.¹²⁰ Another study utilized 4-dimensional MRI by recording the maximum intensity projection from two distinct data, namely, image and feature, within a CNN via SVM classifier.¹²¹ Based on their analysis, the image level exhibited a 91% precision, whereas, the feature level achieved a 93% accuracy.¹²¹ However, their dataset was not generalized and did not include clinically sound information.¹²¹

A multiparametric radiomics model combining DCE- and diffusion-weighted imaging-extracted profiles revealed an optimal AUC (0.85; 95% confidence interval, 0.77, 0.92) and diagnostic precision (81.7%; confidence interval, 73.0, 88.6).¹²² Therefore, radiomics analysis combined with multiparametric MRI ML provides an enhanced assessment of suspicious augmenting breast tumours that are indicated for biopsy on clinical breast MRI.¹²² This facilitates highly precise BC detection while minimizing needless b-BT biopsies.¹²²

Newly developed imaging techniques, such as US, MG, computed tomography (CT), PET and MRI are frequently employed for early BC detection. Among them, however, MRI ranks first in terms of patient prognosis, diagnostic precision, staging and presurgical planning. Furthermore, MRI also exhibits enhanced SEN than MG; and MRI diagnostic results are only minimally impacted by breast density. Hence, MRI is regarded as an essential tool for BC clinical diagnosis.¹²³ A previous investigation achieved an AUC of 0.654 via an ML model that analysed MRI images to delineate between triple-negative breast cancer and remaining subtypes.¹²⁴ The current review extracted and organized the data in tabular form and summarized the application of MRI in breast cancer diagnosis (Table 3).^{109,119,120,123,125–166}

Table 3.

Models, classes and performance for breast magnetic resonance imaging data in selected papers.^{109,119,120,123,125–166}

Paper reference	Models/algorithm	Binary or multiclass	Classes	Accuracy	Other performance evaluation parameters	Anomaly application/task
¹²⁵	ML	Binary	Malignant/benign	81.5%	Specificity = 91.4%	BC detection
¹¹⁹	SVM, DT	Binary	Malignant/benign	86.7%	ROC = 91.1%	BC additional lesion detection
¹²²	SVM	Binary	Malignant/benign	81.7%	AUC = 0.85	BC detection
¹²⁶	SVM, KNN, RF	Binary	Malignant/benign	78.50	AUC = 78.50Sensitivity = 0.923Specificity = 0.722	BC diagnosis
¹²⁷	SVM, LR, NB, KNN	Binary	Malignant/benign	0.93	Sensitivity = 0.85Specificity = 0.89AUC = 90.9%	BC diagnosis
¹²⁸	LR	Binary	Malignant/benign	–	AUC = 95.25	BC diagnosis
¹²³	SVM-RFE, RUSBoost	Binary	Malignant/benign	–	AUC = 0.9617	BC diagnosis
¹²⁹	DCNN, VGGNet	Binary	Malignant/benign		AUC = 0.88	Breast lesion classification
¹³⁰	RF	Binary	Malignant/benign	–	ROC = 0.852	Breast lesions classification
¹³¹	DT	Binary	3+IHC/FISH+	83.9%	Sensitivity = 86.5%Specificity = 80.0%	HER2 expression level
¹³²	SVM, Bayes, KNN, RF, DT	Binary	PCR/nonPCR	–	AUC = 0.879Specificity = 82.19% Sensitivity = 83.57%	BC NACT Efficacy
¹³³	RF, LR, NB	Multiclass	Luminal A/luminal B/HER2-enriched/triple-negative	–	AUC = 0.75	BC phenotype
¹³⁴	SVM	Binary	Malignant/benign	74.1%	AUC = 0.77 ± 0.06	BC diagnosis
¹²⁰	CNN, SVM	Binary	Malignant/benign	–	AUC = 0.87	BC diagnosis
¹³⁵	SVM	Binary	Malignant/benign	–	AUC = 0.87	BC diagnosis
¹³⁶	SVM	Binary	Malignant/benign	–	AUC = 0.925	BC diagnosis
¹³⁷	IsoSVM	Binary	Malignant/benign	_	Sensitivity = 82.5% Specificity = 80.5%AUC = 0.87	BC detection
¹³⁸	U-Net CNN	Binary	Malignant/benign	98.97 ± 0.13	Sensitivity = 96.20 ± 0.82Specificity = 99.54 ± 0.10	BC detection
¹³⁹	LR, ML	Multiclass	TP/FP/negative	–	AUC = 0.50–0.53	Preoperative Breast detection
¹⁴⁰	DCNN, AdaBoost	Multiclass	Malignant/benign/normal	97.2%	Sensitivity = 98.3%Specificity = 96.5%	BC detection
¹⁴¹	KNN, DT, NB, RF, SVM	Binary	Malignant/benign	83.67%	–	BC diagnosis
¹⁴²	RF, LR, GNB, LDA, SVM, MLP	Multiclass	HR+/HER2−/HER2+/TNBC	0.735	AUC = 0.896	BC subtype identification
¹⁴³	DT	Binary	PD-L1+/ PD-L1−	88.2%	Sensitivity = 90.7%Specificity = 85.1%	BC PD-L1 expression status
¹⁴⁴	LR, SVM, RF	Binary	Complete responders/non-complete responders	–	AUC = 0.91–0.98	BC neoadjuvant chemotherapy response
¹⁴⁵	KNN, RF	Binary	Malignant/benign	94.1%	Sensitivity = 99.0%Specificity = 87.8%	Breast tumour classification
¹⁴⁶	RF, SVM, PCA	Binary	Malignant/benign	–	AUC = 0.85	BC diagnosis
¹⁴⁷	SVM	Binary	Malignant/benign	83.2%	–	BC diagnosis
¹⁴⁸	ML	Binary	Malignant/benign	–	–	BC diagnosis
¹⁴⁹	LogitBoost, RF	Binary	luminal A/luminal B	–	Sensitivity = 80.0%	BC prognosis
¹⁵⁰	SVM, NB, LDA, LR	Binary	Malignant/benign	85.9%	Sensitivity = 95.3%Specificity = 71.2%	BC differential diagnosis
¹⁵¹	LDA, SVM	Binary	Malignant/benign	0.84	AUC = 0.88	BC detection
¹⁵²	SVM	Binary	Malignant/benign	–	AUC = 0.90	BC diagnosis
¹⁵³	SVM	Binary	Poor response/excellent response	–	AUC = 0.75	BC neoadjuvant chemotherapy Response
¹⁵⁴	Inception-v3, VGG-16, ResNet50, Inception-ResNet-v2	Multiclass	ER–/+/PR–/+/PAM50 subtypes	0.742	AUC = 0.942AUC = 0.920	BC subtype identification
¹⁵⁵	SVM	Binary	Malignant/benign	88.5%	AUC = 0.96	Breast lesion diagnosis
¹⁰⁹	LR, DT, RF, Bagging	Binary	DCIS/IDC	–	AUC = 0.66–0.78	BC diagnosis
¹⁵⁶	CNN-SVM	Binary	Malignant/benign	95.28%	AUC = 0.974	BC classification
¹⁵⁷	CNN	Binary	Normal/abnormal	88.02%	AUC = 0.880	BC LNM detection
¹⁵⁸	SVM	Binary	Malignant/benign	–	AUC = 0.983	BC diagnosis
¹⁵⁹	NB, RF, AB, DT, KNN, SVM, LDA, LR	Binary	Low/high	–	AUC = 0.79	BC Ki-67 and histological grade detection
¹⁶⁰	SVM, RF, XGB	Binary	Phyllodes tumours/fibroadenomas	93.0%	AUC = 0.97sensitivity = 92.0%specificity = 94.0%	Breast lesion differential diagnosis
¹⁶¹	RF, DT, SVM	Binary	Malignant/benign	0.88	AUC = 0.95	BC diagnosis
¹⁶²	VGG-16, KNN, XGBoost, LightGBM	Binary	Negative/positive	–	AUC = 0.90	BC LNM detection
¹⁶³	RF	Binary	PCR/non-PCR	80.0%	AUC = 0.77	BC neoadjuvant chemotherapy Response
¹⁶⁴	LR	Binary	Negative/positive	–	AUC = 0.76	BC recurrence detection
¹⁶⁵	NB	Binary	Malignant/benign	67%	Sensitivity of 70%Specificity of 66%	BC diagnosis
¹⁶⁶	SVM, RF, DT, NB, LR, LDA, MLP	Binary	Negative/positive	0.82	AUC = 0.85	TNBC TP53 mutation status

ML, machine learning; BC, breast cancer; SVM, support vector machine; DT, decision tree; ROC, receiver operating characteristic; AUC, area under the curve; KNN, k-nearest neighbour; RF, random forests: LR, logistic regression; NB, naïve Bayes; SVM-RFE, support vector machine-based recursive feature elimination; RUSBoost, random undersampling boosting; DCNN, deep convolutional neural networks; VGGNet, visual geometry group network; IHC, immunohistochemistry; FISH, fluorescence in situ hybridization; HER2, human epidermal growth factor receptor 2; PCR, pathological complete response; NACT, neoadjuvant chemotherapy; CNN, convolutional neural network; IsoSVM, isomap support vector machines; U-Net, U-network; TP, true positive; FP, false positive; AdaBoost, adaptive boosting; GNB, Gaussian Naïve Bayes; LDA, linear discriminant analysis; MLP, multilayer perceptron; HR, hormone receptors; TNBC, triple negative breast cancer; PD-L1, programmed death receptor ligand 1; PCA, principal component analysis; VGG-16, visual geometry group-16; ResNet50, residual network 50; ER, oestrogen receptor; PR, progesterone receptor; PAM50, prediction analysis of microarray 50; DCIS, ductal carcinoma in situ; IDC, invasive ductal carcinoma; CNN-SVM, convolutional neural networks-support vector machine; LNM, lymph node metastasis; AB, adaptive boosting; XGB, extreme gradient boosting; XGBoost, extreme gradient boosting; GBM, gradient boosting machine.

Others

X-ray mammography is often employed for BC identification. However, it is a highly invasive procedure because X-rays damage tissues and quite frequently fails to determine the tumour size. Lately, thermography has emerged as a safer non-invasive approach, with no contact imaging. This process does not involve ionizing radiation, venous access or other invasive protocols. Thermography used human body-emitted infrared electromagnetic radiation that is picked up by a thermographic camera for analysis by a CAD system. However, for widespread application, it is imperative to enhance the accuracy of these new tools. ML has successfully enhanced diagnostic precision while minimizing the presence of false positives and false negatives while analysing breast thermograms.

Various forms of research have examined thermography-based BC identification. This current review will present recent significant publications. A study published in 2023 reported a new AI- and thermography-based CAD system that assists radiologists in accurately diagnosing breast diseases.¹⁶⁷ The procedure for this new tool is as follows: using the U-net model, an intersection over an 89.03% union is achieved.¹⁶⁷ The segmented thermograms then undergo textural assessment and vascular network analysis to isolate significant profiles.¹⁶⁷ Subsequently, via implementation of the supervised learning algorithm-based classifiers and usage of the retrieved profiles, the normal versus abnormal thermograms are identified.¹⁶⁷ This process of BC identification was further confirmed as highly effective, revealing optimal stratification while using SVM, with a 94.4% accuracy, 96.2% precision, 86.7% recall, 91.2% F1-score and 98.3% true negative rate.¹⁶⁷ Another study conducted a binary stratification of m-BT and b-BT using breast thermographic images.¹⁶⁸ The authors employed 94 images (320 × 240) with biopsy-confirmed diagnoses, among which, 60 exhibited m-BT and 34 exhibited b-BT.¹⁶⁸ The authors used three distinct image analytical tools, namely, blinded screening mode (SBS), clinical assessment and ANN.¹⁶⁸ The first method provided a risk score ranging from 0 (minimum risk) to 7 (very high risk).¹⁶⁸ The remaining methods yielded a binary result, identifying whether a given lesion was m-BT or b-BT.¹⁶⁸ The analyses revealed that the ANN method excelled over the others, with a 81.8% accuracy, relative to 66.7% for SBS and 71.4% for clinical analysis.¹⁶⁸ Another study reported using CNN that included data augmentation and a fine-tuning optimization algorithm and a an automatic BC diagnosis.¹⁶⁹ These received a 92% accuracy and demonstrated that data augmentation considerably enhanced tumour stratification in breast thermography, particularly when data were scarce.¹⁶⁹

Furthermore, among studies examining thermographic images for BC classification, one demonstrated exhibited an SEN of 0.812 and an SPC of 0.882.¹⁷⁰ Additionally, screening tomosynthesis also garnered much attention owing to its enhancement of cancer detection rates, along with diminished false positive rates.¹⁷¹ Similarly, a study evaluated screen-identification approaches for tomosynthetic single-reading versus double-reading mammograms, and demonstrated an 8.2 versus 6.3 cancer detection rate per 1000 screens.¹⁷²

Additionally, there is also BT identification using highly advanced microwave systems, which facilitate a much safer, non-ionizing approach to delineate between healthy and non-healthy tissues, based on individual dielectric profiles. The present microwave breast imaging research can be categorized as follows: microwave tomography (MT) and ultra-wideband (UWB) radar techniques. MT utilizes antennas with a matching liquid, whereas UWB employs 60 antennas with a matching liquid.¹⁷³

More recently, ML approaches have been extensively examined for detecting BTs. In addition, DL methods have also been extensively examined. Currently, for BT microwave imaging, ML and DL have been used to analyse microwave datasets from numerical simulations or phantoms measurements.¹⁷⁴ Research evaluated for the first-time cancer detection using UWB enhanced by ML and conventional breast examinations.¹⁷⁴ The authors demonstrated that the SVM quadratic kernel classified breast information with 98% precision.¹⁷⁴ The current review extracted and organized the data in tabular form and summarized the application of CT, breast thermal imaging, PET and microwave imaging technology in the diagnosis of breast cancer (Table 4).^{92,109,147,149,151,167,174–187}

Table 4.

Models, classes and performance for breast thermographic techniques, positron emission tomography and other combined examination data in selected papers.^{92,109,147,149,151,167,174–187}

Paper reference	Relevant examinations	Models/algorithm	Binary or multiclass	Classes	Accuracy	Other performance evaluation parameters	Anomaly application/task
¹⁷⁵	CBCT	RF, KNN, BPN, SVM	Binary	Malignant/benign	–	AUC = 0.91Sensitivity = 0.85Specificity = 0.82	BC detection
¹⁷⁶	PET	SVM	Binary	Malignant/benign	0.85	AUC = 0.89Sensitivity = 0.94Specificity = 0.77	BC diagnosis
¹⁷⁷	PET	CNN, MLP	Binary	Malignant/benign	–	AUC = 0.947	BC diagnosis
¹⁴⁹	PET	LogitBoost, RF	Binary	Luminal A/Luminal B	–	AUC = 75%Sensitivity = 88.0%	BC subtypes
¹⁷⁸	PET	MultiResUnet3D	Binary	Negative/positive	–	AUC = 0.775	BC NAC response
¹⁷⁹	Thermography	CNN, ELM, SVM, RF, MLP	Binary	Malignant/benign	0.8843	AUC = 0.9675	BC detection
¹⁸⁰	Microwave	RF	Binary	Malignant/benign	0.87	–	BC diagnosis
¹⁸¹	PS-OCT	SVM	Multiclass	Malignant/fibroadipose/stroma	93.5%	–	Breast tissue types
¹⁷⁴	UWB	MLP-NN, SVM, KNN	Binary	Lesion-containing/lesion-free	98%	Sensitivity = 0.97Specificity = 0.99	Breast lesion detection
¹⁶⁷	Thermography	DNN, SVM	Binary	Normal/ abnormal	94.4%	F1-score = 91.2%	BC Detection
¹⁸²	PAT	SVM, AlexNet, GoogleNet	Binary	Malignant/benign	0.9118	AUC = 0.8143Sensitivity = 0.8571	BC diagnosis
¹⁸³	DCE-MRI, MGs	RBFNN, KNN, SVM, RF	Binary	Malignant/benign	–	AUC = 99.10%	BC detection
¹⁴⁷	DCE-MRI, MG	SVM	Binary	Malignant/benign	83.3%	Specificity of 82.1%	BC diagnosis
¹⁰⁹	MG, US, MRI	LR, DT, RF, Bagging	Binary	DCIS/ IDC	–	AUC = 0.66–0.78	BC diagnosis
¹⁸⁴	MG	RF, SVM, LMT, NB, KNN	Binary	Malignant/benign	–	Wins = 83%	BC detection
¹⁸⁵	Pathological images	deep learning, CNN, InceptionResNetV2	Binary	Cancerous/non-cancerous	91%	–	BC risk detection
¹⁸⁶	US	XGBoost Tree, LR	Binary	Malignant/benign	–	AUC = 0.90Sensitivity = 100%	Breast masses classification
⁹²	US	SL, ML	Binary	Malignant/benign	95.7%	Sensitivity = 0.912Specificity = 0.966	BC diagnosis
¹⁸⁷	X-rays, US, MI, PET, EIT	PSOWNN	Multiclass	Malignant/benign/normal	98.6%	Specificity = 98.8%	BC detection
¹⁵¹	CEM, DCE-MRI	LDA, SVM	Binary	Malignant/benign	0.84	AUC = 0.88	BC detection

CBCT, cone-beam computed tomography; RF, random forests; KNN, k-nearest neighbour; BPN, back propagation neural networks; SVM, support vector machine; AUC, area under the curve; BC, breast cancer; PET, positron emission tomography; CNN, convolutional neural network; MLP, multilayer perceptron; NAC, neoadjuvant chemotherapy; ELM, extreme learning machines; PS-OCT, polarization-sensitive optical coherence tomography; UWB, ultra-wideband; MLP-NN, multi-layer perceptron neural network; DNN, deep neural network; PAT, photoacoustic tomography; DCE-MRI, dynamic contrast enhancement-magnetic resonance imaging; MGs, mammographic images; RBFNN, radial basis function neural network; MG, mammography; US, ultrasonography; MRI, magnetic resonance imaging; LR, logistic regression; DT, decision tree; DCIS, ductal carcinoma in situ; IDC, invasive ductal carcinoma; LMT, logistic model trees; NB, naïve Bayes; XGBoost, extreme gradient boosting; SL, supervised learning; ML, machine learning; MI, microwave imaging; EIT, electrical impedance tomography; PSOWNN, particle swarm optimized wavelet neural network; CEM, contrast-enhanced mammography.

Conclusion

In conclusion, the use of ML techniques for BC imaging has the potential to improve the accuracy and efficiency of diagnosis, classification and prediction, thereby improving patient outcomes and reducing healthcare costs. ML still has its limitations, both in terms of imaging and pathological diagnosis, and ML cannot make a diagnosis of untrained diseases. In addition, ML requires a large amount of training data support, but because of research confidentiality or patient privacy protection, the effectiveness, safety and universality of data are the key issues troubling the clinical application of ML. However, further research is needed to address the challenges and limitations of these techniques, and to develop standardized protocols and benchmarks for evaluating their performance.

Supplemental Material

sj-pdf-1-imr-10.1177_03000605241237867 - Supplemental material for Machine learning and new insights for breast cancer diagnosis

Supplemental material, sj-pdf-1-imr-10.1177_03000605241237867 for Machine learning and new insights for breast cancer diagnosis by Ya Guo, Heng Zhang, Leilei Yuan, Weidong Chen, Haibo Zhao, Qing-Qing Yu and Wenjie Shi in Journal of International Medical Research

Footnotes

Author contributions

Wenjie Shi and Qing-Qing Yu conceived of the presented idea, drafted and reviewed the relevant literature. Leilei Yuan, Weidong Chen and Haibo Zhao contributed to the design of this study. Ya Guo and Heng Zhang contributed to data collection. All authors have read and agreed to the published version of the manuscript.

Declaration of conflicting interest

The authors declare that there are no conflicts of interest.

Funding

This work was supported by the Key R&D Program of Jining (no. 2023YXNS051).

ORCID iDs

Qing-Qing Yu

Wenjie Shi

References

Sung

Ferlay

Siegel

, et al. Global Cancer Statistics 2020: GLOBOCAN Estimates of Incidence and Mortality Worldwide for 36 Cancers in 185 Countries. CA Cancer J Clin 2021; 71: 209–249.

Anaya-Isaza

Mera-Jimenez

Cabrera-Chavarro

, et al. Comparison of Current Deep Convolutional Neural Networks for the Segmentation of Breast Masses in Mammograms. IEEE Access 2021; 9: 152206–152225.

Akin

Brennan

Dershaw

, et al. Advances in oncologic imaging: update on 5 common cancers. CA Cancer J Clin 2012; 62: 364–393.

Mao

Yin

Wang

, et al. Added Value of Radiomics on Mammography for Breast Cancer Diagnosis: A Feasibility Study. J Am Coll Radiol 2019; 16: 485–491.

Sakai

Onishi

Matsui

, et al. A method for the automated classification of benign and malignant masses on digital breast tomosynthesis images using machine learning and radiomic features. Radiol Phys Technol 2020; 13: 27–36.

Zhu

AlZoubi

Jassim

, et al. A generic deep learning framework to classify thyroid and breast lesions in ultrasound images. Ultrasonics 2021; 110: 106300.

Fanizzi

Losurdo

Basile

TMA

, et al. Fully Automated Support System for Diagnosis of Breast Cancer in Contrast-Enhanced Spectral Mammography Images. J Clin Med 2019; 8: 891.

Wan

Wong

, et al. Evaluation of the performance of traditional machine learning algorithms, convolutional neural network and AutoML Vision in ultrasound breast lesions classification: a comparative study. Quant Imaging Med Surg 2021; 11: 1381–1393.

Liu

Tong

Wan

, et al. Identification and diagnosis of mammographic malignant architectural distortion using a deep learning based mask regional convolutional neural network. Front Oncol 2023; 13: 1119743.

10.

Esteva

Kuprel

Novoa

, et al. Dermatologist-level classification of skin cancer with deep neural networks. Nature 2017; 542: 115–118.

11.

Franck

Snoeckx

Spinhoven

, et al. Pulmonary Nodule Detection in Chest Ct Using a Deep Learning-Based Reconstruction Algorithm. Radiat Prot Dosimetry 2021; 195: 158–163.

12.

Geras

Mann

Moy

Artificial Intelligence for Mammography and Digital Breast Tomosynthesis: Current Concepts and Future Perspectives. Radiology 2019; 293: 246–259.

13.

Cole

Zhang

Marques

, et al. Impact of computer-aided detection systems on radiologist accuracy with digital mammography. AJR Am J Roentgenol 2014; 203: 909–916.

14.

Lehman

Wellman

Buist

, et al. Diagnostic Accuracy of Digital Screening Mammography With and Without Computer-Aided Detection. JAMA Intern Med 2015; 175: 1828–1837.

15.

Agarwal

Diaz

Llado

, et al. Automatic mass detection in mammograms using deep convolutional neural networks. J Med Imaging (Bellingham) 2019; 6: 031409.

16.

Hamidinekoo

Denton

Rampun

, et al. Deep learning in mammography and breast histology, an overview and future trends. Med Image Anal 2018; 47: 45–67.

17.

Kim

Han

, et al. Applying Data-driven Imaging Biomarker in Mammography for Breast Cancer Screening: Preliminary Study. Sci Rep 2018; 8: 2762.

18.

Aboutalib

Mohamed

Berg

, et al. Deep Learning to Distinguish Recalled but Benign Mammography Images in Breast Cancer Screening. Clin Cancer Res 2018; 24: 5902–5909.

19.

Rodriguez-Ruiz

Lang

Gubern-Merida

, et al. Stand-Alone Artificial Intelligence for Breast Cancer Detection in Mammography: Comparison With 101 Radiologists. J Natl Cancer Inst 2019; 111: 916–922.

20.

Rodriguez-Ruiz

Krupinski

Mordang

, et al. Detection of Breast Cancer with Mammography: Effect of an Artificial Intelligence Support System. Radiology 2019; 290: 305–314.

21.

Chougrad

Zouaki

Alheyane

Deep Convolutional Neural Networks for breast cancer screening. Comput Methods Programs Biomed 2018; 157: 19–30.

22.

Hosny

Schabath

, et al. Artificial intelligence in cancer imaging: Clinical challenges and applications. CA Cancer J Clin 2019; 69: 127–157.

23.

Saba

Khan

Islam

, et al. Cloud-based decision support system for the detection and classification of malignant cells in breast cancer using breast cytology images. Microsc Res Tech 2019; 82: 775–785.

24.

Al-Fahaidy

FAK

Al-Fuhaidi

Al-Darouby

, et al. A Diagnostic Model of Breast Cancer Based on Digital Mammogram Images Using Machine Learning Techniques. Applied Computational Intelligence and Soft Computing 2022; 2022: 1–17.

25.

Al-Masni

Al-Antari

Park

, et al. Simultaneous detection and classification of breast masses in digital mammograms via a deep learning YOLO-based CAD system. Comput Methods Programs Biomed 2018; 157: 85–94.

26.

Sannasi Chakravarthy

Rajaguru

Automatic Detection and Classification of Mammograms Using Improved Extreme Learning Machine with Deep Learning. IRBM 2022; 43: 49–61.

27.

Ragab

Sharkas

Marshall

, et al. Breast cancer detection using deep convolutional neural networks and support vector machines. PeerJ 2019; 7: e6201.

28.

Assiri

Nazir

Velastin

SA.

Breast Tumor Classification Using an Ensemble Machine Learning Method. J Imaging 2020; 6: 39.

29.

Tan

Zheng

Ramalingam

, et al. Prediction of Near-term Breast Cancer Risk Based on Bilateral Mammographic Feature Asymmetry. Acad Radiol 2013; 20: 1542–1550.

30.

Mookiah

MRK

Banerjee

Chakraborty

, et al. Statistical analysis of mammographic features and its classification using support vector machine. Expert Systems with Applications 2010; 37: 470–478.

31.

de Lima

da Silva

Dos Santos

WP.

Detection and classification of masses in mammographic images in a multi-kernel approach. Comput Methods Programs Biomed 2016; 134: 11–29.

32.

Wang

Yang

Cai

, et al. Discrimination of Breast Cancer with Microcalcifications on Mammography by Deep Learning. Sci Rep 2016; 6: 27327.

33.

Becker

Marcon

Ghafoor

, et al. Deep Learning in Mammography Diagnostic Accuracy of a Multipurpose Image Analysis Software in the Detection of Breast Cancer. Invest Radiol 2017; 52: 434–440.

34.

Samala

Chan

Hadjiiski

, et al. Multi-task transfer learning deep convolutional neural network: application to computer-aided diagnosis of breast cancer on mammograms. Phys Med Biol 2017; 62: 8894–8908.

35.

Bart

Hegde

Deep Synthesis of Realistic Medical Images: A Novel Tool in Clinical Research and Training. Front Neuroinform 2018; 12: 82.

36.

Celaya Padilla

Guzman Valdivia

Galvan Tejada

, et al. Contralateral asymmetry for breast cancer detection: A CADx approach. Journal of Applied Biomedicine 2018; 38: 115–125.

37.

Patel

Ranjbar

, et al. Computer-aided diagnosis of contrast-enhanced spectral mammography: A feasibility study. Eur J Radiol 2018; 98: 207–213.

38.

Qasem

Abdullah

SNHS

Sahran

, et al. Heterogeneous Ensemble Pruning based on Bee Algorithm for Mammogram Classification. International Journal of Advanced Computer Science and Applications 2018; 9: 231–239.

39.

Perez-Benito

Signol

Perez-Cortes

, et al. Global parenchymal texture features based on histograms of oriented gradients improve cancer development risk estimation from healthy breasts. Comput Methods Programs Biomed 2019; 177: 123–132.

40.

Shen

Margolies

Rothstein

, et al. Deep Learning to Improve Breast Cancer Detection on Screening Mammography. Sci Rep 2019; 9: 12495.

41.

Agarwal

Díaz

Yap

, et al. Deep learning for mass detection in Full Field Digital Mammograms. Comput Biol Med 2020; 121: 103774.

42.

Arora

Rai

Raman

Deep feature-based automatic classification of mammograms. Med Biol Eng Comput 2020; 58: 1199–1211.

43.

Samala

Chan

Hadjiiski

, et al. Generalization error analysis for deep convolutional neural network with transfer learning in breast cancer diagnosis. Phys Med Biol 2020; 65: 105002.

44.

Schaffter

Buist

DSM

Lee

, et al. Evaluation of Combined Artificial Intelligence and Radiologist Assessment to Interpret Screening Mammograms. JAMA Netw Open 2020; 3: e200265.

45.

Shu

Zhang

Wang

, et al. Deep Neural Networks With Region-Based Pooling Structures for Mammographic Image Classification. IEEE Trans Med Imaging 2020; 39: 2246–2255.

46.

Stelzer

Steding

Raudner

, et al. Combined texture analysis and machine learning in suspicious calcifications detected by mammography: Potential to avoid unnecessary stereotactical biopsies. Eur J Radiol 2020; 132: 109309.

47.

Alhussan

Abdel Samee

Ghoneim

, et al. Evaluating Deep and Statistical Machine Learning Models in the Classification of Breast Cancer from Digital Mammograms. International Journal of Advanced Computer Science and Applications 2021; 12: 304–313.

48.

Alì

D’Amico

Interlenghi

, et al. A Decision Support System Based on BI-RADS and Radiomic Classifiers to Reduce False Positive Breast Calcifications at Digital Breast Tomosynthesis: A Preliminary Study. Applied Science 2021; 11: 2503.

49.

Chouhan

Khan

Shah

, et al. Deep convolutional neural network and emotional learning based breast cancer detection using digital mammography. Comput Biol Med 2021; 132: 104318.

50.

Saeed Darweesh

Adel

Anwar

, et al. Early breast cancer diagnostics based on hierarchical machine learning classification for mammography images. Cogent Engineering 2021; 8: 1.

51.

Jamal

Ishak

Abdel-Khalek

Tumor edge detection in mammography images using quantum and machine learning approaches. Neural Computing and Applications 2021; 33: 1–12.

52.

Kazemi

Shiri

Sheikhahmadi

, et al. A new parallel deep learning algorithm for breast cancer classification. International Journal of Nonlinear Analysis and Applications 2021; 12: 1269–1282.

53.

Mahmood

Pei

, et al. An Automatic Detection and Localization of Mammographic Microcalcifications ROI with Multi-Scale Features Using the Radiomics Analysis Approach. Cancers (Basel) 2021; 13: 5916.

54.

Samala

Chan

Hadjiiski

, et al. Risks of feature leakage and sample size dependencies in deep feature extraction for breast mass classification. Med Phys 2021; 48: 2827–2837.

55.

Tran

Sadeghi-Naini

, et al. Computational Radiology in Breast Cancer Screening and Diagnosis Using Artificial Intelligence. Can Assoc Radiol J 2021; 72: 98–108.

56.

Zebari

Ibrahim

Zeebaree

, et al. Breast Cancer Detection Using Mammogram Images with Improved Multi-Fractal Dimension Approach and Feature Fusion. Appl Sci 2021; 11: 12122.

57.

Almalki

Khalid

Alduraibi

, et al. LBP–Bilateral Based Feature Fusion for Breast Cancer Diagnosis. Computers Materials & Continua 2022; 73: 4103–4121.

58.

Alruwaili

Gouda

Automated Breast Cancer Detection Models Based on Transfer Learning. Sensors (Basel) 2022; 22: 876.

59.

Altameem

Mahanty

Poonia

, et al. Breast Cancer Detection in Mammography Images Using Deep Convolutional Neural Networks and Fuzzy Ensemble Modeling Techniques. Diagnostics (Basel) 2022; 12: 1812.

60.

Alyami

Sadad

Rehman

, et al. Cloud Computing-Based Framework for Breast Tumor Image Classification Using Fusion of AlexNet and GLCM Texture Features with Ensemble Multi-Kernel Support Vector Machine (MK-SVM). Comput Intell Neurosci 2022; 2022: 7403302.

61.

Annamalai

Chinnasamy

Pandian

A Hybrid Model Particle Swarm Optimization Based Mammogram Classification Using Kernel Support Vector Machine. Traitement du Signal 2022; 39: 915–922.

62.

Chen

Zhang

Zhou

, et al. Diagnosis of architectural distortion on digital breast tomosynthesis using radiomics and deep learning. Front Oncol 2022; 12: 991892.

63.

Dafni Rose

VijayaKumar

Singh

, et al. Computer-aided diagnosis for breast cancer detection and classification using optimal region growing segmentation with MobileNet model. Concurrent Engineering 2022; 30: 181–189.

64.

Das

Neog

, et al. Breast cancer detection: Shallow convolutional neural network against deep convolutional neural networks based approach. Front Genet 2022; 13: 1097207.

65.

Houssein

Emam

Ali

AA.

An optimized deep learning architecture for breast cancer diagnosis based on improved marine predators algorithm. Neural Comput Appl 2022; 34: 18015–18033.

66.

Islam

Jones

Faiz

, et al. Improving Performance of Breast Lesion Classification Using a ResNet50 Model Optimized with a Novel Attention Mechanism. Tomography 2022; 8: 2411–2425.

67.

Khan

Saba

Sadad

, et al. Identification of Anomalies in Mammograms through Internet of Medical Things (IoMT) Diagnosis System. Comput Intell Neurosci 2022; 2022: 1100775.

68.

Mansour

Althobaiti

MM.

Cognitive Computing-Based Mammographic Image Classification on an Internet of Medical. Computers Materials & Continua 2022; 72: 3945–3959.

69.

Thirunavukkarasu

Breast Cancer Semantic Segmentation for Accurate Breast Cancer Detection with an Ensemble Deep Neural Network. Neural Processing Letters 2022; 54: 5185–5198.

70.

Ramesh

Sasikala

Swaminathan

, et al. Segmentation and classification of breast cancer using novel deep learning architecture. Neural Computing and Applications 2022; 34: 1–13.

71.

Rehman

Pei

, et al. Architectural Distortion-Based Digital Mammograms Classification Using Depth Wise Convolutional Neural Network. Biology (Basel) 2022; 11: 15.

72.

Samee

Alhussan

Ghoneim

, et al. A Hybrid Deep Transfer Learning of CNN-Based LR-PCA for Breast Lesion Diagnosis via Medical Breast Mammograms. Sensors (Basel) 2022; 22: 4938.

73.

Wang

Sun

, et al. Diagnostic performance of perilesional radiomics analysis of contrast-enhanced mammography for the differentiation of benign and malignant breast lesions. Eur Radiol 2022; 32: 639–649.

74.

Zahedi

Moridani

MK.

Classification of Breast Cancer Tumors Using Mammography Images Processing Based on Machine Learning. International Journal of Online & Biomedical Engineering 2022; 18: 31.

75.

Casal-Guisande

Alvarez-Pazo

Cerqueiro-Pequeno

, et al. Proposal and Definition of an Intelligent Clinical Decision Support System Applied to the Screening and Early Diagnosis of Breast Cancer. Cancers (Basel) 2023; 15: 1711.

76.

Haraguchi

Goto

Furuya

, et al. Use of machine learning with two-dimensional synthetic mammography for axillary lymph node metastasis prediction in breast cancer: a preliminary study. Transl Cancer Res 2023; 12: 1232–1240.

77.

Jones

Sadeghipour

Chen

, et al. A multi-stage fusion framework to classify breast lesions using deep learning and radiomics features computed from four-view mammograms. Med Phys 2023; 50: 7670–7683.

78.

Ali

, Sahib

Ali

MA.

Investigation of Early-Stage Breast Cancer Detection using Quantum Neural Network. International Journal of Online and Biomedical Engineering 2023; 19: 61–81.

79.

Sarvestani

Jamali

Taghizadeh

, et al. A novel machine learning approach on texture analysis for automatic breast microcalcification diagnosis classification of mammogram images. J Cancer Res Clin Oncol 2023; 149: 6151–6170.

80.

Barinov

Jairaj

Becker

, et al. Impact of Data Presentation on Physician Performance Utilizing Artificial Intelligence-Based Computer-Aided Diagnosis and Decision Support Systems. J Digit Imaging 2019; 32: 408–416.

81.

Bae

Han

Koo

, et al. Characteristics of breast cancers detected by ultrasound screening in women with negative mammograms. Cancer Sci 2011; 102: 1862–1867.

82.

Romeo

Pinker

Helbich

TH.

Chapter 10 – Breast imaging. Clinical PET/MRI 2023; 245–266.

83.

Singh

Verma

Panigrahi

, et al. Integrating radiologist feedback with computer aided diagnostic systems for breast cancer risk prediction in ultrasonic images: An experimental investigation in machine learning paradigm. Expert Systems with Applications 2017; 90: 209–223.

84.

Venkatesh

Levenback

Sultan

, et al. Going beyond a First Reader: A Machine Learning Methodology for Optimizing Cost and Performance in Breast Ultrasound Diagnosis. Ultrasound Med Biol 2015; 41: 3148–3162.

85.

Song

Venkatesh

Conant

, et al. Comparative analysis of logistic regression and artificial neural network for computer-aided diagnosis of breast masses. Acad Radiol 2005; 12: 487–495.

86.

Moustafa

Cary

Sultan

, et al. Color Doppler Ultrasound Improves Machine Learning Diagnosis of Breast Cancer. Diagnostics (Basel) 2020; 10: 631.

87.

Zhang

Song

Xiao

, et al. Dual-mode artificially-intelligent diagnosis of breast tumours in shear-wave elastography and B-mode ultrasound using deep polynomial networks. Med Eng Phys 2019; 64: 1–6.

88.

Liao

Hao

, et al. Automatic Identification of Breast Ultrasound Image Based on Supervised Block-Based Region Segmentation Algorithm and Features Combination Migration Deep Learning Model. IEEE J Biomed Health Inform 2020; 24: 984–993.

89.

Lee

Chang

Chou

, et al. Fully automated lesion segmentation and visualization in automated whole breast ultrasound (ABUS) images. Quant Imaging Med Surg 2020; 10: 568–584.

90.

Sultan

Tian

, et al. Machine learning for diagnostic ultrasound of triple-negative breast cancer. Breast Cancer Res Treat 2019; 173: 365–373.

91.

Zhang

, et al. Computer-aided diagnosis of breast cancer in ultrasonography images by deep learning. Neurocomputing 2022; 472: 152–165.

92.

Raj

Rahman

SMK

Anand

Preliminary evaluation of differentiation of benign and malignant breast tumors using non-invasive diagnostic modalities. Biomedical Research 2016; 27: 596–603.

93.

Guo

Wang

, et al. Automatic tumor segmentation in breast ultrasound images using a dilated fully convolutional network combined with an active contour model. Med Phys 2019; 46: 215–228.

94.

Marcon

Ciritsis

Rossi

, et al. Diagnostic performance of machine learning applied to texture analysis-derived features for breast lesion characterisation at automated breast ultrasound: a pilot study. Eur Radiol Exp 2019; 3: 44.

95.

Casagrande

Fabris

Girometti

Beyond kappa: an informational index for diagnostic agreement in dichotomous and multivalue ordered-categorical ratings. Med Biol Eng Comput 2020; 58: 3089–3099.

96.

Wang

Choi

, et al. Breast cancer classification in automated breast ultrasound using multiview convolutional neural network with transfer learning. Ultrasound Med Biol 2020; 46: 1119–1132.

97.

Fei

Zhou

Han

, et al. Doubly supervised parameter transfer classifier for diagnosis of breast cancer with imbalanced ultrasound imaging modalities. Pattern Recognition 2021; 120: 108139.

98.

Huo

Tan

Wang

, et al. Machine Learning Models to Improve the Differentiation Between Benign and Malignant Breast Lesions on Ultrasound: A Multicenter External Validation Study. Cancer Manag Res 2021; 13: 3367–3379.

99.

Liu

Parker

Jung

SH.

Design and Analysis Methods for Trials with AI-Based Diagnostic Devices for Breast Cancer. J Pers Med 2021: 11: 1150.

100.

Mishra

Roy

Bandyopadhyay

, et al. Breast ultrasound tumour classification: A Machine Learning-Radiomics based approach. Expert Systems 2021; 38: e12713.

101.

Baek

O'Connell

Parker

KJ.

Improving breast cancer diagnosis by incorporating raw ultrasound parameters into machine learning. Mach Learn Sci Technol 2022; 3: 045013.

102.

Hamyoon

Chan

Mohammadi

, et al. Artificial intelligence, BI-RADS evaluation and morphometry: A novel combination to diagnose breast cancer using ultrasonography, results from multi-center cohorts. Eur J Radiol 2022; 157: 110591.

103.

Hoffmann

Reich

Skerl

Evaluating different combination methods to analyse ultrasound and shear wave elastography images automatically through discriminative convolutional neural network in breast cancer imaging. Int J Comput Assist Radiol Surg 2022; 17: 2231–2237.

104.

Homayoun

Chan

Kuzan

, et al. Applications of machine-learning algorithms for prediction of benign and malignant breast lesions using ultrasound radiomics signatures: A multi-center study. Biocybernetics and Biomedical Engineering 2022; 42: 921–933.

105.

Huang

Chen

, et al. Preoperative Non-Invasive Prediction of Breast Cancer Molecular Subtypes With a Deep Convolutional Neural Network on Ultrasound Images. Front Oncol 2022; 12: 848790.

106.

Liu

Guo

Liu

, et al. MfdcModel: A Novel Classification Model for Classification of Benign and Malignant Breast Tumors in Ultrasound Images. Electronics 2022; 11: 2583.

107.

Magnuska

Theek

Darguzyte

, et al. Influence of the Computer-Aided Decision Support System Design on Ultrasound-Based Breast Cancer Classification. Cancers (Basel) 2022; 14: 277.

108.

O'Connell

Bartolotta

Orlando

, et al. Diagnostic Performance of An Artificial Intelligence System in Breast Ultrasound. J Ultrasound Med 2022; 41: 97–105.

109.

Park

Kim

Han

, et al. Ductal carcinoma in situ: a risk prediction model for the underestimation of invasive breast cancer. NPJ Breast Cancer 2022; 8: 8.

110.

Pfob

Sidey-Gibbons

Barr

, et al. Intelligent multi-modal shear wave elastography to reduce unnecessary biopsies in breast cancer diagnosis (INSPiRED 002): a retrospective, international, multicentre analysis. Eur J Cancer 2022; 177: 1–14.

111.

Pham

Faust

Koh

JEW

, et al. Fusion of B-mode and shear wave elastography ultrasound features for automated detection of axillary lymph node metastasis in breast carcinoma. Expert Systems 2022; 39: e12947.

112.

Ragab

Albukhari

Alyami

, et al. Ensemble Deep-Learning-Enabled Clinical Decision Support System for Breast Cancer Diagnosis and Classification on Ultrasound Images. Biology (Basel) 2022; 11: 439.

113.

Tang

Liang

Tao

, et al. Machine learning-based diagnostic evaluation of shear-wave elastography in Bi-rads category 4 breast cancer screening: a multicenter, retrospective study. Quant Imaging Med Surg 2022; 12: 1223–1234.

114.

Vigil

Barry

Amini

, et al. Dual-Intended Deep Learning Model for Breast Cancer Diagnosis in Ultrasound Imaging. Cancers (Basel) 2022; 14: 2663.

115.

Wang

Chen

, et al. Multi-region radiomics for artificially intelligent diagnosis of breast cancer using multimodal ultrasound. Comput Biol Med 2022; 149: 105920.

116.

Zhang

Shi

Yin

, et al. A machine learning model based on ultrasound image features to assess the risk of sentinel lymph node metastasis in breast cancer patients: Applications of scikit-learn and SHAP. Front Oncol 2022; 12: 944569.

117.

Zhu

Lin

, et al. Ultrasound-based radiomics analysis for differentiating benign and malignant breast lesions: From static images to CEUS video analysis. Front Oncol 2022; 12: 951973.

118.

Abbasian Ardakani

Mohammadi

Mirza-Aghazadeh-Attari

, et al. An open-access breast lesion ultrasound image database: Applicable in artificial intelligence studies. Comput Biol Med 2023; 152: 106438.

119.

Lee

Nguyen

, et al. Classification of MR-Detected Additional Lesions in Patients With Breast Cancer Using a Combination of Radiomics Analysis and Machine Learning. Front Oncol 2021; 11: 744460.

120.

Whitney

Giger

ML.

A deep learning methodology for improved breast cancer diagnosis using multiparametric MRI. Sci Rep 2020; 10: 10536.

121.

Whitney

, et al. Improved Classification of Benign and Malignant Breast Lesions Using Deep Feature Maximum Intensity Projection MRI in Breast Cancer Diagnosis Using Dynamic Contrast-enhanced MRI. Radiol Artif Intell 2021; 3: e200159.

122.

Daimiel Naranjo

Gibbs

Reiner

, et al. Radiomics and Machine Learning with Multiparametric Breast MRI for Improved Diagnostic Accuracy in Breast Cancer Diagnosis. Diagnostics (Basel) 2021; 11: 919.

123.

Chu

A novel computer-aided diagnosis system for breast MRI based on feature selection and ensemble learning. Comput Biol Med 2017; 83: 157–165.

124.

Saha

Harowicz

Grimm

, et al. A machine learning approach to radiogenomics of breast cancer: a study of 922 subjects and 529 DCE-MRI features. Br J Cancer 2018; 119: 508–516.

125.

Lo Gullo

Daimiel

Rossi Saccarelli

, et al. Improved characterization of sub-centimeter enhancing breast masses on MRI with radiomics and machine learning in BRCA mutation carriers. Eur Radiol 2020; 30: 6721–6731.

126.

Cai

Liu

Peng

, et al. Diagnostic assessment by dynamic contrast-enhanced and diffusion-weighted magnetic resonance in differentiation of breast lesions under different imaging protocols. BMC Cancer 2014; 14: 366.

127.

Cai

Peng

, et al. Diagnosis of Breast Masses from Dynamic Contrast-Enhanced and Diffusion-Weighted MR: A Machine Learning Approach. PloS One 2014; 9: e87387.

128.

Bickelhaupt

Paech

Kickingereder

, et al. Prediction of malignancy by a radiomic signature from contrast agent-free diffusion MRI in suspicious breast lesions found on screening mammography. J Magn Reson Imaging 2017; 46: 604–616.

129.

Antropova

Abe

Giger

ML.

Use of clinical MRI maximum intensity projections for improved breast lesion classification with deep convolutional neural networks. J Med Imaging (Bellingham) 2018; 5: 014503.

130.

Dalmiş

Gubern-Mérida

Vreemann

, et al. Artificial Intelligence-Based Classification of Breast Lesions Imaged With a Multiparametric Breast MRI Protocol With Ultrafast DCE-MRI, T2, and DWI. Invest Radiol 2019; 54: 325–332.

131.

Bitencourt

AGV

Gibbs

Rossi Saccarelli

, et al. MRI-based machine learning radiomics can predict HER2 expression level and pathologic response after neoadjuvant therapy in HER2 overexpressing breast cancer. EBioMedicine 2020; 61: 103042.

132.

Chen

Shu

, et al. Machine Learning-Based Radiomics Nomogram Using Magnetic Resonance Images for Prediction of Neoadjuvant Chemotherapy Efficacy in Breast Cancer Patients. Front Oncol 2020; 10: 1410.

133.

Demircioglu

Grueneisen

Ingenwerth

, et al. A rapid volume of interest-based approach of radiomics analysis of breast MRI for tumor decoding and phenotyping of breast cancer. PLoS One 2020; 15: e0234871.

134.

Hao

Gong

Wang

, et al. Application of MRI Radiomics-Based Machine Learning Model to Improve Contralateral BI-RADS 4 Lesion Assessment. Front Oncol 2020; 10: 531476.

135.

Whitney

Giger

ML.

Radiomics methodology for breast cancer diagnosis using multiparametric magnetic resonance imaging. J Med Imaging (Bellingham) 2020; 7: 044502.

136.

Jiang

Yin

JD.

Performance evaluation of texture analysis based on kinetic parametric maps from breast DCE-MRI in classifying benign from malignant lesions. J Surg Oncol 2020; 121: 1181–1190.

137.

Parekh

Jacobs

MA.

Multiparametric radiomics methods for breast cancer tissue characterization using radiological imaging. Breast Cancer Res Treat 2020; 180: 407–421.

138.

Piantadosi

Sansone

Fusco

, et al. Multi-planar 3D breast segmentation in MRI via deep convolutional neural networks. Artif Intell Med 2020; 103: 101781.

139.

Rahbar

Hippe

Alaa

, et al. The Value of Patient and Tumor Factors in Predicting Preoperative Breast MRI Outcomes. Radiol Imaging Cancer 2020; 2: e190099.

140.

Zheng

Lin

Gao

, et al. Deep Learning Assisted Efficient AdaBoost Algorithm for Breast Cancer Detection and Early Diagnosis. IEEE Access 2020; 8: 96946–96954.

141.

Avuclu

A new data augmentation method to use in machine learning algorithms using statistical measurements. Measurement 2021; 180: 109577.

142.

Huang

Wei

, et al. Multi-Parametric MRI-Based Radiomics Models for Predicting Molecular Subtype and Androgen Receptor Expression in Breast Cancer. Front Oncol 2021; 11: 706733.

143.

Lo Gullo

Wen

Reiner

, et al. Assessing PD-L1 Expression Status Using Radiomic Features from Contrast-Enhanced Breast MRI in Breast Cancer Patients: Initial Results. Cancers (Basel) 2021; 13: 6273.

144.

Montemezzi

Benetti

Bisighin

, et al. 3T DCE-MRI Radiomics Improves Predictive Models of Complete Response to Neoadjuvant Chemotherapy in Breast Cancer. Front Oncol 2021; 11: 630780.

145.

Sayed

AM.

Machine Learning Augmented Breast Tumors Classification using Magnetic Resonance Imaging Histograms. International Journal of Advanced Computer Science and Applications 2021; 12: 1–9.

146.

Sun

Jiao

Zhu

, et al. Radiomics-based machine learning analysis and characterization of breast lesions with multiparametric diffusion-weighted MR. J Transl Med 2021; 19: 443.

147.

Zhao

Chen

Zhang

, et al. Diagnosis of Breast Cancer Using Radiomics Models Built Based on Dynamic Contrast Enhanced MRI Combined With Mammography. Front Oncol 2021; 11: 774248.

148.

Baltzer

PAT

Krug

Dietzel

Evidence-Based and Structured Diagnosis in Breast MRI using the Kaiser Score. Rofo 2022; 194: 1216–1228.

149.

Castaldo

Garbino

Cavaliere

, et al. A Complex Radiomic Signature in Luminal Breast Cancer from a Weighted Statistical Framework: A Pilot Study. Diagnostics (Basel) 2022; 12: 499.

150.

Chen

Zhao

Liu

, et al. Assessment of breast lesions by the Kaiser score for differential diagnosis on MRI: the added value of ADC and machine learning modeling. Eur Radiol 2022; 32: 6608–6618.

151.

Fusco

Di Bernardo

Piccirillo

, et al. Radiomic and Artificial Intelligence Analysis with Textural Metrics Extracted by Contrast-Enhanced Mammography and Dynamic Contrast Magnetic Resonance Imaging to Detect Breast Malignant Lesions. Curr Oncol 2022; 29: 1947–1966.

152.

Whitney

, et al. Impact of continuous learning on diagnostic breast MRI AI: evaluation on an independent clinical dataset. J Med Imaging (Bellingham) 2022; 9: 034502.

153.

McAnena

Moloney

Browne

, et al. A radiomic model to classify response to neoadjuvant chemotherapy in breast cancer. BMC Med Imaging 2022; 22: 225.

154.

Ming

Zhu

, et al. Predicting hormone receptors and PAM50 subtypes of breast cancer from multi-scale lesion images of DCE-MRI with transfer learning technique. Comput Biol Med 2022; 150: 106147.

155.

Daimiel Naranjo

Gibbs

Reiner

, et al. Breast Lesion Classification with Multiparametric Breast MRI Using Radiomics and Machine Learning: A Comparison with Radiologists' Performance. Cancers (Basel) 2022; 14: 1743.

156.

Rashid

Ibrikci

Paydaş

, et al. Analysis of breast cancer classification robustness with radiomics feature extraction and deep learning techniques. Expert Systems 2022; 39: e13018.

157.

Ren

Lin

Huang

, et al. Convolutional Neural Network of Multiparametric MRI Accurately Detects Axillary Lymph Node Metastasis in Breast Cancer Patients With Pre Neoadjuvant Chemotherapy. Clin Breast Cancer 2022; 22: 170–177.

158.

Romeo

Clauser

Rasul

, et al. AI-enhanced simultaneous multiparametric F-18-FDG PET/MRI for accurate breast cancer diagnosis. Eur J Nucl Med Mol Imaging 2022; 49: 596–608.

159.

Song

Cho

, et al. Machine learning with multiparametric breast MRI for prediction of Ki-67 and histologic grade in early-stage luminal breast cancer. Eur Radiol 2022; 32: 853–863.

160.

Tsuchiya

Masui

Terauchi

, et al. MRI-based radiomics analysis for differentiating phyllodes tumors of the breast from fibroadenomas. Eur Radiol 2022; 32: 4090–4100.

161.

Vamvakas

Tsivaka

Logothetis

, et al. Breast Cancer Classification on Multiparametric MRI – Increased Performance of Boosting Ensemble Methods. Technol Cancer Res Treat 2022; 21: 15330338221087828.

162.

Wang

Zhan

, et al. A nomogram based on radiomics signature and deep-learning signature for preoperative prediction of axillary lymph node metastasis in breast cancer. Front Oncol 2022; 12: 940655.

163.

Yoshida

Kawashima

Kannon

, et al. Prediction of pathological complete response to neoadjuvant chemotherapy in breast cancer using radiomics of pretreatment dynamic contrast-enhanced MRI. Magn Reson Imaging 2022; 92: 19–25.

164.

Chiacchiaretta

Mastrodicasa

Chiarelli

, et al. MRI-Based Radiomics Approach Predicts Tumor Recurrence in ER+/HER2- Early Breast Cancer Patients. J Digit Imaging 2023; 36: 1071–1080.

165.

Stogiannos

Bougias

Georgiadou

, et al. Analysis of radiomic features derived from post-contrast T1-weighted images and apparent diffusion coefficient (ADC) maps for breast lesion evaluation: A retrospective study. Radiography (Lond) 2023; 29: 355–361.

166.

Sun

Zhu

Chai

, et al. Multimodality MRI radiomics analysis of TP53 mutations in triple negative breast cancer. Front Oncol 2023; 13: 1153261.

167.

Chebbah

Ouslim

Benabid

New computer aided diagnostic system using deep neural network and SVM to detect breast cancer in thermography. Quantitative Infrared Thermography Journal 2023; 20: 62–77.

168.

Arora

Martins

Ruggerio

, et al. Effectiveness of a noninvasive digital infrared thermal imaging system in the detection of breast cancer. Am J Surg 2008; 196: 523–526.

169.

Zuluaga-Gomez

Al Masry

Benaggoune

, et al. A CNN-based methodology for breast cancer diagnosis using thermal images. Computer Methods in Biomechanics and Biomedical Engineering: Imaging & Visualization 2021; 9: 131–145.

170.

EYK

Kee

Acharya

UR.

Advanced technique in breast thermography analysis. 2005 IEEE Engineering in Medicine and Biology 27th Annual Conference, Shanghai, China, 2005, pp. 710–713, doi: 10.1109/IEMBS.2005.1616512.

171.

Skaane

Bandos

Gullien

, et al. Prospective trial comparing full-field digital mammography (FFDM) versus combined FFDM and tomosynthesis in a population-based screening programme using independent double reading with arbitration. Eur Radiol 2013; 23: 2061–2071.

172.

Houssami

Bernardi

Pellegrini

, et al. Breast cancer detection using single-reading of breast tomosynthesis (3D-mammography) compared to double-reading of 2D-mammography: Evidence from a population-based trial. Cancer Epidemiol 2017; 47: 94–99.

173.

Preece

Craddock

Shere

, et al. MARIA M4: clinical evaluation of a prototype ultrawideband radar scanner for breast cancer detection. J Med Imaging (Bellingham) 2016; 3: 033502.

174.

Rana

Dey

Tiberi

, et al. Machine Learning Approaches for Automated Lesion Detection in Microwave Breast Imaging Clinical Data. Sci Rep 2019; 9: 10510.

175.

Uhlig

Kunze

, et al. Novel Breast Imaging and Machine Learning: Predicting Breast Lesion Malignancy at Cone-Beam CT Using Machine Learning Techniques. AJR Am J Roentgenol 2018; 211: W123–W131.

176.

Satoh

Tamada

Omiya

, et al. Diagnostic Performance of the Support Vector Machine Model for Breast Cancer on Ring-Shaped Dedicated Breast Positron Emission Tomography Images. J Comput Assist Tomogr 2020; 44: 413–418.

177.

Caballo

Hernandez

Lyu

, et al. Computer-aided diagnosis of masses in breast computed tomography imaging: deep learning model with combined handcrafted and convolutional radiomic features. J Med Imaging (Bellingham) 2021; 8: 024501.

178.

Hian

Kumaran

, et al. Multi-center evaluation of artificial intelligent imaging and clinical models for predicting neoadjuvant chemotherapy response in breast cancer. Breast Cancer Res Treat 2022; 193: 121–138.

179.

Macedo

Santana

dos Santos

, et al. Breast cancer diagnosis using thermal image analysis: A data-driven approach based on swarm intelligence and supervised learning for optimized feature selection. Applied Soft Computing 2021; 109: 107533.

180.

Oliveira

Godinho

O’Halloran

, et al. Diagnosing Breast Cancer with Microwave Technology: Remaining Challenges and Potential Solutions with Machine Learning. Diagnostics (Basel) 2018; 8: 36.

181.

Zhu

Wang

Marjanovic

, et al. Differentiation of breast tissue types for surgical margin assessment using machine learning and polarization-sensitive optical coherence tomography. Biomed Opt Express 2021; 12: 3021–3036.

182.

Zhang

Chen

Zhou

, et al. Photoacoustic Image Classification and Segmentation of Breast Cancer: A Feasibility Study. IEEE Access 2019; 7: 5457–5466.

183.

Mokni

Gargouri

Damak

, et al. An automatic Computer-Aided Diagnosis system based on the Multimodal fusion of Breast Cancer (MF-CAD). Biomedical Signal Processing and Control 2021; 69: 102914.

184.

Moura

Guevara López

MA.

An evaluation of image descriptors combined with clinical data for breast cancer diagnosis. Int J Comput Assist Radiol Surg 2013; 8: 561–574.

185.

Humayun

Khalil

Almuayqil

, et al. Framework for Detecting Breast Cancer Risk Presence Using Deep Learning. Electronics 2023; 12: 403.

186.

Pfob

Sidey-Gibbons

Barr

, et al. The importance of multi-modal imaging and clinical information for humans and AI-based algorithms to classify breast masses (INSPiRED 003): an international, multicenter analysis. Eur Radiol 2022; 32: 4101–4115.

187.

Nomani

Ansari

Nasirpour

, et al. PSOWNNs-CNN: A Computational Radiology for Breast Cancer Diagnosis Improvement Based on Image Processing Using Machine Learning Methods. Comput Intell Neurosci 2022; 2022: 5667264.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.36 MB