Classification of Breast Cancer using Ensemble Filter Feature Selection with Triplet Attention Based Efficient Net Classifier

Breast cancer remains a major global health concern, particularly among women, where early detection is crucial for effective treatment and increased survival rates. Manual diagnosis through histopathological examination is often time-consuming, prone to human error, and requires expert interpretation. In response to this challenge, we developed a deep learning-based system to automate the detection and classification of breast cancer using microscopic biopsy images, aimed at enhancing diagnostic accuracy and speed in clinical workflows. Our project employs convolutional neural networks (CNNs) to classify histopathological images into three distinct categories: normal, benign, and malignant. We utilized a publicly available dataset containing high-resolution microscopic images and implemented preprocessing techniques to improve image quality, normalize input dimensions, and optimize feature extraction. The model was developed and trained using TensorFlow with GPU support, allowing for efficient computation and faster training cycles. Key Words: Breast cancer, Cancer detection, Cancer classification, Histopathological images, microscopic images, Digital pathology, medical imaging

Convolutional Neural Network (CNN) models are a type of deep learning architecture introduced to achieve the correct classification of breast cancer. This paper has a two-fold purpose. The first aim is to investigate the various deep learning models in classifying breast cancer histopathology images. This study identified the most accurate models in terms of the binary, four, and eight classifications of breast cancer histopathology image databases. The different accuracy scores obtained for the deep learning models on the same database showed that other factors such as pre-processing, data augmentation, and transfer learning methods can impact the ability of the models to achieve higher accuracy. The second purpose of our manuscript is to investigate the latest models that have no or limited examination done in previous studies. The models like ResNeXt, Dual Path Net, SENet, and NASNet had been identified with the most cutting-edge results for the ImageNet database. These models were examined for the binary, and eight classifications on BreakHis, a breast cancer histopathology image database. Furthermore, the BACH database was used to investigate these models for four classifications. Then, these models were compared with the previous studies to find and propose the most state-of-the-art models for each classification. Since the Inception-ResNet-V2 architecture achieved the best results for binary and eight classifications, we have examined this model in our study as well to provide a better comparison result. In short, this paper provides an extensive evaluation and discussion about the experimental settings for each study that had been conducted on the breast cancer histopathology images.

Breast cancer is a common and fatal disease among women worldwide. Therefore, the early and precise diagnosis of breast cancer plays a pivotal role to improve the prognosis of patients with this disease. Several studies have developed automated techniques using different medical imaging modalities to predict breast cancer development. However, few review studies are available to recapitulate the existing literature on breast cancer classification. These studies provide an overview of the classification, segmentation, or grading of many cancer types, including breast cancer, by using traditional machine learning approaches through hand-engineered features. This review focuses on breast cancer classification by using medical imaging multimodalities through state-of-the-art artificial deep neural network approaches. It is anticipated to maximize the procedural decision analysis in five aspects, such as types of imaging modalities, datasets and their categories, pre-processing techniques, types of deep neural network, and performance metrics used for breast cancer classification. Forty-nine journal and conference publications from eight academic repositories were methodically selected and carefully reviewed from the perspective of the five aforementioned aspects. In addition, this study provided quantitative, qualitative, and critical analyses of the five aspects. This review showed that mammograms and histopathologic images were mostly used to classify breast cancer. Moreover, about 55% of the selected studies used public datasets, and the remaining used exclusive datasets. Several studies employed augmentation, scaling, and image normalization pre-processing techniques to minimize inconsistencies in breast cancer images. Several types of shallow and deep neural network architecture were employed to classify breast cancer using images. The convolutional neural network was utilized frequently to construct an effective breast cancer classification model. Some of the selected studies employed a pre-trained network or developed new deep neural networks to classify breast cancer. Most of the selected studies used accuracy and area-under-the-curve metrics followed by sensitivity, precision, and F-measure metrics to evaluate the performance of the developed breast cancer classification models. Finally, this review presented 10 open research challenges for future scholars who are interested to develop breast cancer classification models through various imaging modalities. This review could serve as a valuable resource for beginners on medical image classification and for advanced scientists focusing on deep learning-based breast cancer classification through different medical imaging modalities.

In medical imaging, precise recognition of Breast Cancer (BC) is a challenge due to the complications of breast tissues. Histopathological detection is still considered the standard in BC detection. Still, the dramatic increase in workload and the complexity of histopathological image (HPI) make this task labor-intensive and dependent on the pathologist, making the advance of automated and precise HPI analysis techniques needed. Due to the automated feature extraction capability, Deep Learning (DL) methods have been effectively used in different sectors, particularly in the medical imaging sector. This study develops the future search algorithm with a DL-based breast cancer detection and classification (FSADL-BCDC) method. The FSADL-BCDC technique examines HPIs to detect and classify BC. To achieve this, the FSADL-BCDC technique implements Wiener Filtering (WF)-based preprocessing to eliminate the noise in the images. Additionally, the FSADL-BCDC uses the ResNeXt method for feature extraction with a Future Search Algorithm (FSA)-based tuning procedure. For BCDC, the FSADL-BCDC technique employs a Hybrid Convolutional Neural Network along with the Long Short-Term Memory (HCNN-LSTM) approach. Finally, the Sunflower Optimization (SFO) approach adjusts the hyperparameter values of the HCNN-LSTM. The outcomes of the FSADL-BCDC are inspected on a standard medical image dataset. Extensive relational studies highlighted the improved performance of the FSADL-BCDC approach in comparison with known methods by exhibiting an output of 96.94% and 98.69% under diverse datasets.

Breast cancer encompasses various subtypes with distinct prognoses, necessitating accurate stratification methods. Current techniques rely on quantifying gene expression in limited subsets. Given the complexity of breast tissues, effective detection and classification of breast cancer is crucial in medical imaging. This study introduces a novel method, MPa-DCAE, which uses a multi-patch-based deep convolutional auto-encoder (DCAE) framework combined with VGG19 to detect and classify breast cancer in histopathology images. The proposed MPa-DCAE model leverages the hierarchical feature extraction capabilities of VGG19 within a DCAE framework, designed to capture intricate patterns in histopathology images. By using a multi-patch approach, regions of interest are extracted from pathology images to facilitate localized feature learning, enhancing the model's discriminatory power. The auto-encoder component enables unsupervised feature learning, increasing resilience and adaptability to variations in image features. Experiments were conducted at various magnifications on the CBIS-DDSM and MIAS datasets to validate model performance. Experimental results demonstrated that the MPa-DCAE model outperformed existing methods. For the CBIS-DDSM dataset, the model achieved a precision of 97.96%, a recall of 94.85%, and an accuracy of 98.36%. For the MIAS dataset, it achieved a precision of 97.99%, a recall of 97.2%, and an accuracy of 98.95%. These results highlight the model's robustness and potential for clinical application in computer-assisted diagnosis. The MPa-DCAE model, integrating VGG19 and DCAE, proves to be an effective, automated approach for diagnosing breast cancer. Its high accuracy and generalizability make it a promising tool for clinical practice, potentially improving patient care in histopathology-based breast cancer diagnosis.

Globally, breast cancer represents the leading cancer type, with millions of women impacted annually. The success of breast cancer treatment relies heavily on timely detection and precise tumor classification. The classification of breast cancer has gained considerable importance in Deep Learning (DL) and medical research with the development of medical imaging techniques, like histopathological imaging. Many existing DL schemes suffer from overfitting and endure difficulties in effectively mining the key features from high-resolution images with subtle variations. Hence, the Xception Convolutional Deep Maxout Network (Xcov-DMN) is developed to classify breast cancer. At the initial stage, the Mean-Shift Filter is applied to the input histopathological image. Following this, the White Blood Cell Network (WBC-Net) is employed for blood cell segmentation with the Balanced Cross-Entropy (BCE) and Focal Loss for ensuring precise segmentation. Next, Colored Histograms, shape features, Haralick Texture Features, and Complete Local Binary Pattern (CLBP) features are excerpted. Consequently, the developed Xcov-DMN is utilized to classify breast cancer. Xcov-DMN is the combination of the Deep Maxout Network (DMN), Fractional Calculus (FC), and Xception Convolutional Neural Network (XCovNet). Moreover, with learning data at 90%, the Xcov-DMN achieved the highest accuracy of 92.755%, True Negative Rate (TNR) of 91.977%, and True Positive Rate (TPR) of 94.765%.

In this study, a multiscale framework combining the Lifting Wavelet Transform (LWT) and multi-path Convolutional Neural Network (CNN) is proposed to enhance the analysis of histopathological breast cancer images. LWT is employed due to its capability to obtain multi-resolution features that effectively retain key textural details. In particular, a multi-path CNN facilitates the concurrent processing of features at multiple levels of wavelet decompositions, thereby preserving diagnostic information that sole-path CNN models may lose. The approach was evaluated on the BreakHis dataset, which contains 7,638 rated images at varying magnifications, achieving 99.34% test accuracy by combining magnification levels using a Haar wavelet filter. Results obtained from comparative analyses against general CNN models confirm that the new strategy performs better, thereby validating the efficacy of integrating wavelet-based textural extraction and state-of-the-art CNN models for early-stage breast cancer diagnosis.

Breast cancer is considered the second most common reason for death among women. The gold standard for detecting breast cancer is the visual interpretation of histopathological images, but it is a complicated process that takes years of experience and a lot of skills of the pathologists. Sometimes, the limitations of the visual interpretation and the lack of experience result in the failure of diagnosing breast cancer. So, the computer-aided diagnosis (CAD) system can be taken into consideration as a helping tool to reduce the error of diagnosis of breast cancer. In this paper, a novel approach based on convolutional neural networks is introduced to classify breast cancer from the histopathological images of the breast tissues. In the clinical process of breast cancer diagnosis, pathologists examine the histopathological images of the breast tissue at different magnification levels. In this research, a single convolutional neural networks model is used to take the input of the same image with four different magnification levels parallelly. Our proposed approach outperformed existing state-of-the-art approaches by a substantial margin.

Continuous blood pressure (BP) provides essential information for monitoring one's health condition. However, BP is currently monitored using uncomfortable cuff-based devices, which does not support continuous BP monitoring. This paper aims to introduce a blood pressure monitoring algorithm based on only photoplethysmography (PPG) signals using the deep neural network (DNN). The PPG signals are obtained from 125 unique subjects with 218 records and filtered using signal processing algorithms to reduce the effects of noise, such as baseline wandering, and motion artifacts. The proposed algorithm is based on pulse wave analysis of PPG signals, extracted various domain features from PPG signals, and mapped them to BP values. Four feature selection methods are applied and yielded four feature subsets. Therefore, an ensemble feature selection technique is proposed to obtain the optimal feature set based on major voting scores from four feature subsets. DNN models, along with the ensemble feature selection technique, outperformed in estimating the systolic blood pressure (SBP) and diastolic blood pressure (DBP) compared to previously reported approaches that rely only on the PPG signal. The coefficient of determination ( ) and mean absolute error (MAE) of the proposed algorithm are 0.962 and 2.480 mmHg, respectively, for SBP and 0.955 and 1.499 mmHg, respectively, for DBP. The proposed approach meets the Advancement of Medical Instrumentation standard for SBP and DBP estimations. Additionally, according to the British Hypertension Society standard, the results attained Grade A for both SBP and DBP estimations. It concludes that BP can be estimated more accurately using the optimal feature set and DNN models. The proposed algorithm has the potential ability to facilitate mobile healthcare devices to monitor continuous BP.

Background and Objectives: The classification of breast cancer is performed based on its histological subtypes using the degree of differentiation. However, there have been low levels of intra- and inter-observer agreement in the process. The use of convolutional neural networks (CNNs) in the field of radiology has shown potential in categorizing medical images, including the histological classification of malignant neoplasms. Materials and Methods: This study aimed to use CNNs to develop an automated approach to aid in the histological classification of breast cancer, with a focus on improving accuracy, reproducibility, and reducing subjectivity and bias. The study identified regions of interest (ROIs), filtered images with low representation of tumor cells, and trained the CNN to classify the images. Results: The major contribution of this research was the application of CNNs as a machine learning technique for histologically classifying breast cancer using medical images. The study resulted in the development of a low-cost, portable, and easy-to-use AI model that can be used by healthcare professionals in remote areas. Conclusions: This study aimed to use artificial neural networks to improve the accuracy and reproducibility of the process of histologically classifying breast cancer and reduce the subjectivity and bias that can be introduced by human observers. The results showed the potential for using CNNs in the development of an automated approach for the histological classification of breast cancer.

ความเป็นมา การตอบสนองทางพยาธิวิทยาภายหลังการได้รับยาเคมีบำบัดก่อนผ่าตัดมีความสัมพันธ์กับผลลัพธ์ทางคลินิก ปัจจุบันไม่มีปัจจัยใดที่สามารถทำนายการตอบสนองต่อยาเคมีบำบัดในมะเร็งเต้านมชนิดมีตัวรับฮอร์โมนและไม่มีตัวรับเฮอร์ทูได้อย่างแม่นยำ เช่นเดียวกับ Ki-67 การแสดงออกของโปรตีน MCM2 ซึ่งมีหน้าที่ควบคุมการสังเคราะห์ดีเอ็นเอในระดับสูงบ่งชี้ถึงการแบ่งตัวของเซลล์ที่รวดเร็ว และมีความสัมพันธ์กับการพยากรณ์โรคที่ไม่ดีในมะเร็งหลายชนิดรวมถึงมะเร็งเต้านม การศึกษานี้มีวัตถุประสงค์เพื่อศึกษาการแสดงออกของ MCM2 ในชิ้นเนื้อก่อนได้รับยาเคมีบำบัดเพื่อทำนายการตอบสนองทางพยาธิวิทยาแบบ RCB 0/I ในมะเร็งเต้านมระยะต้นชนิดมีตัวรับฮอร์โมนและไม่มีตัวรับเฮอร์ทู วิธีการศึกษา ทบทวนเวชระเบียนอิเล็กทรอนิกส์ของผู้ป่วยมะเร็งเต้านมระยะต้นชนิดมีตัวรับฮอร์โมนและไม่มีตัวรับเฮอร์ทูที่ได้รับยาเคมีบำบัดก่อนผ่าตัดในโรงพยาบาลจุฬาลงกรณ์ ระหว่าง 1 มกราคม 2558 - 31 ธันวาคม 2563 นำชิ้นเนื้อก่อนได้รับยาเคมีบำบัดตรวจการแสดงออกของ MCM2 ด้วยการย้อมอิมมูโนฮิสโตเคมีและทบทวนลักณะทางพยาธิวิทยาเพื่อคำนวน RCB index ในชิ้นเนื้อหลังได้รับยาเคมีบำบัด คำนวนการติดสีของ MCM2 ด้วยโปรแกรมวิเคราะห์ภาพ Aperio Imagescope ผลการวิจัย ผู้ป่วยทั้งหมด 88 คน มีค่ามัธยฐานอายุ 49 ปี (พิสัย 27-80 ปี) 61.4% เป็นวัยก่อนหมดประจำเดือน มีระยะโรคมะเร็งทางคลินิก cT4 และ cN1 40% และ 66.7% ตามลำดับ เป็นมะเร็งเต้านมชนิด invasive ductal carcinoma 91.1% มีระดับทางพยาธิวิทยาระดับที่ 2 58.9% มีการแสดงออกของ ER >10% 92.2% และสถานะ PR เป็นลบ 22.2% ในชิ้นเนื้อมะเร็งเต้านมทั้งหมด 72 ชิ้นที่สามารถย้อม MCM2 และประเมิน RCB index ได้ พบว่าการแสดงออกของ MCM2 >=40% มีความสัมพันธ์กับการตอบสนองทางพยาธิวิทยาแบบ RCB 0/I อย่างมีนัยสำคัญทางสถิติจากการวิเคราะห์แบบ univariate analysis (OR = 18.33 ช่วงความเชื่อมั่น 95% = 1.88-178.98 p-value = 0.012) ปัจจัยทางพยาธิวิทยาคลินิกที่พบว่ามีความสัมพันธ์กับ MCM2 >=40% คือ ระดับของพยาธิวิทยาระดับที่ 3 การแสดงออกของ ER <=10% สถานะ PR เป็นลบ และ Ki-67 >=20% แต่ในการวิเคราะห์แบบ multivariate analysis ไม่มีปัจจัยทางพยาธิวิทยาคลินิกใดสัมพันธ์กับการตอบสนองทางพยาธิวิทยาแบบ RCB 0/I ลักษณะทางอิมมูโนฮิสโตเคมีก่อนและหลังให้ยาเคมีบำบัดมีลักษณะโดยรวมใกล้เคียงกัน ค่ามัธยฐานของการปลอดโรค และอัตราการปลอดโรคที่ 3 ปี ไม่มีความแตกต่างกันอย่างมีนัยสำคัญทางสถิติระหว่างผู้ป่วยที่มีการตอบสนองทางพยาธิวิทยาแบบ RCB 0/I และ RCB II/III สรุปผลการวิจัย การแสดงออกของ MCM2 >=40% ด้วยการตรวจทางอิมมูโนฮิสโตเคมี ในชิ้นเนื้อก่อนได้รับยาเคมีบำบัดของผู้ป่วยมะเร็งเต้านมระยะต้นชนิดมีตัวรับฮอร์โมนและไม่มีตัวรับเฮอร์ทู มีความสัมพันธ์กับการตอบสนองแบบ RCB 0/I การตรวจการแสดงออกของ MCM2 อาจช่วยทำนายประโยชน์ของยาเคมีบำบัดก่อนผ่าตัดในผู้ป่วยกลุ่มนี้ได้ อย่างไรก็ตามควรมีการศึกษาเพิ่มเติมในประชากรที่มีขนาดใหญ่ขึ้นเพื่อยืนยันผลของการศึกษานี้

Deep and machine learning techniques for medical imaging-based breast cancer: A comprehensive review

Background: The role of HER2 expression in HER2 negative breast cancer is evolving. Whether HER2 expression has an impact on ER positive, HER2 negative early-stage breast cancer in unclear. Oncotype DX assay Recurrence Score comprises of standardized quantitative HER2 mRNA expression levels by RT-PCR. The objective of this study was to investigate the impact of Oncotype HER2 score on tumor characteristics and outcome. Methods: All women assigned to the Clalit Health Services registry between 2006 to 2011 with ER positive, HER2 negative early-stage breast cancer whose tumor were sent for oncotype were identified. Patients in which HER2 score was reported in the oncotype results were included. Differences in age, tumor characteristics and chemotherapy administration between lower HER2 score (H1), i.e. HER2 score ≤ 9.1, to higher HER2 score (H2), i.e. 9.2-10.7, were analyzed. 5-year Kaplan–Meier for distant recurrence free survival (DRFS) and overall survival (OS) were estimated and differences between H1 and H2 were evaluated. Analyses were repeated separately for node negative and for node positive (N mic and 1-3 lymph nodes) disease. Results: A total of 1535 patients were included, 948 patients with node negative disease and 587 with node positive disease. Of these, 814 and 721 patients were categorized to the H1 and H2 groups, respectively. Median follow-up was 5.4 years. The distribution of oncotype RS was different, with significantly lower proportion of patients with RS between 0-10 and higher proportion of RS above 25 in the H1 group compared to the H2 groups. This difference was identified both for patients with node negative and node positive disease (Table). The distributions of age, tumor size and histological subtype were comparable between the H1 and the H2 groups, both for node negative and node positive disease. Within each RS group, administration of chemotherapy was comparable between H1 and H2 groups, regardless of nodal involvement. DRFS and OS were not associated with HER2 score. For RS 0-25, node negative disease, 5-year DRFS was 96.2% in H1 and 96.4% in H2 groups (p=0.92) and 5-year OS was 97.2% in H1 and 98.1% in H2 groups (p=0.39). For RS >25, node negative disease, 5-years DRFS was 89% in H1 and 90.8% in H2 groups (p=0.33) and 5-year OS was 90.6% in H1 and 90.8% in H2 groups (p=0.7). Similarly, for node positive disease, DRFS and OS were comparable between H1 and H2 group in all subgroups. Conclusions: Low HER2 mRNA expression by Oncotype RT-PCR assay was associated with significantly high Oncotype RS. Patients' outcome was not directly affected by HER2 score and was determined by RS. Further study is desired to clarify the role of HER2 expression in early stage, luminal breast cancer. Table. Comparison of age and tumor characteristics between low (H1) and high (H2) HER2 score Citation Format: Hadar Goldvaser, Raz Mutai, Salomon Stemmer, Iryna Kuchuk, Margarita Tokar, Shani Paluch-Shimon, Rinat Yerushalmi, Karen Drumea, Ella Evron, Amir Sonnenblick, Einav Gal-Yam, Gil Bar Sela, Ayelet Shai, Rotem Merose, Avital Bareket-Samish, Lior Soussan-Gutman. The impact of Human Epidermal growth factor Receptor 2 (HER2) expression by RT-PCR on tumor characteristics and outcome in estrogen receptor (ER) positive, HER2 negative early-stage breast cancer [abstract]. In: Proceedings of the 2023 San Antonio Breast Cancer Symposium; 2023 Dec 5-9; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2024;84(9 Suppl):Abstract nr PO5-01-08.

Early detection and precise diagnosis of breast cancer (BC) plays an essential part in enhancing the diagnosis and improving the breast cancer survival rate of patients from 30 to 50%. Through the advances of technology in healthcare, deep learning takes a significant role in handling and inspecting a great number of X-ray, MRI, CTR images. The aim of this study is to propose a deep learning model (BCCNN) to detect and classify breast cancers into eight classes: benign adenosis (BA), benign fibroadenoma (BF), benign phyllodes tumor (BPT), benign tubular adenoma (BTA), malignant ductal carcinoma (MDC), malignant lobular carcinoma (MLC), malignant mucinous carcinoma (MMC), and malignant papillary carcinoma (MPC). Breast cancer MRI images were classified into BA, BF, BPT, BTA, MDC, MLC, MMC, and MPC using a proposed Deep Learning model with additional 5 fine-tuned Deep learning models consisting of Xception, InceptionV3, VGG16, MobileNet and ResNet50 trained on ImageNet database. The dataset was collected from Kaggle depository for breast cancer detection and classification. That Dataset was boosted using GAN technique. The images in the dataset have 4 magnifications (40X, 100X, 200X, 400X, and Complete Dataset). Thus we evaluated the proposed Deep Learning model and 5 pre-trained models using each dataset individually. That means we carried out a total of 30 experiments. The measurement that was used in the evaluation of all models includes: F1-score, recall, precision, accuracy. The classification F1-score accuracies of Xception, InceptionV3, ResNet50, VGG16, MobileNet, and Proposed Model (BCCNN) were 97.54%, 95.33%, 98.14%, 97.67%, 93.98%, and 98.28%, respectively. Dataset Boosting, preprocessing and balancing played a good role in enhancing the detection and classification of breast cancer of the proposed model (BCCNN) and the fine-tuned pre-trained models' accuracies greatly. The best accuracies were attained when the 400X magnification of the MRI images due to their high images resolution.

As one of the most common malignant tumors among women worldwide, breast cancer requires early diagnosis and accurate classification to significantly improve patient survival rates. Conventional imaging techniques like mammography, ultrasound, and magnetic resonance imaging (MRI) play a pivotal role in breast cancer screening. Nonetheless, they are constrained by relatively low specificity and a significant dependence on the expertise of medical professionals. In recent years, machine learning and deep learning techniques have provided new approaches for the intelligent diagnosis of breast cancer by extracting high-dimensional features from medical images. This study delves into the pathological aspects, imaging technologies, and the implementation of machine learning algorithms in the context of breast cancer. It conducts a comprehensive review of the diagnostic criteria for non-invasive, early-stage invasive, and fully invasive breast cancers, while also evaluating the strengths and weaknesses of various imaging modalities, including mammography, ultrasound, MRI, and nuclear medicine imaging. The limitations of conventional imaging methods in subtype differentiation are also discussed. Furthermore, by integrating radiomics and deep learning models such as convolutional neural networks (CNN) and random forests, the study evaluates the performance of intelligent diagnostic systems in breast cancer classification. Clinical cases and publicly available datasets were used as data sources. The results show that combining multimodal imaging features with machine learning algorithms significantly improves diagnostic accuracy, achieving an area under the curve (AUC) of 0.922. This research provides theoretical support and technical references for the precise diagnosis and treatment of breast cancer. Future work should focus on enhancing model generalizability and conducting multi-center clinical validation.