Accuracy Analysis of Deep Learning Methods in Breast Cancer Classification: A Structured Review

Diagnostics, Journal Year: 2023, Volume and Issue: 13(4), P. 683 - 683

Published: Feb. 11, 2023

Breast cancer is diagnosed using histopathological imaging. This task extremely time-consuming due to high image complexity and volume. However, it important facilitate the early detection of breast for medical intervention. Deep learning (DL) has become popular in imaging solutions demonstrated various levels performance diagnosing cancerous images. Nonetheless, achieving precision while minimizing overfitting remains a significant challenge classification solutions. The handling imbalanced data incorrect labeling further concern. Additional methods, such as pre-processing, ensemble, normalization techniques, have been established enhance characteristics. These methods could influence be used overcome balancing issues. Hence, developing more sophisticated DL variant improve accuracy reducing overfitting. Technological advancements fueled automated diagnosis growth recent years. paper reviewed studies on capability classify images, objective this study was systematically review analyze current research Additionally, literature from Scopus Web Science (WOS) indexes reviewed. assessed approaches applications papers published up until November 2022. findings suggest that especially convolution neural networks their hybrids, are most cutting-edge currently use. To find new technique, necessary first survey landscape existing hybrid conduct comparisons case studies.

Language: Английский

---

Color-CADx: a deep learning approach for colorectal cancer classification through triple convolutional neural networks and discrete cosine transform

Scientific Reports, Journal Year: 2024, Volume and Issue: 14(1)

Published: March 22, 2024

Abstract Colorectal cancer (CRC) exhibits a significant death rate that consistently impacts human lives worldwide. Histopathological examination is the standard method for CRC diagnosis. However, it complicated, time-consuming, and subjective. Computer-aided diagnostic (CAD) systems using digital pathology can help pathologists diagnose faster more accurately than manual histopathology examinations. Deep learning algorithms especially convolutional neural networks (CNNs) are advocated diagnosis of CRC. Nevertheless, most previous CAD obtained features from one CNN, these huge dimension. Also, they relied on spatial information only to achieve classification. In this paper, system proposed called “Color-CADx” recognition. Different CNNs namely ResNet50, DenseNet201, AlexNet used end-to-end classification at different training–testing ratios. Moreover, extracted reduced discrete cosine transform (DCT). DCT also utilized acquire spectral representation. Afterward, further select set deep features. Furthermore, coefficients in step concatenated analysis variance (ANOVA) feature selection approach applied choose Finally, machine classifiers employed Two publicly available datasets were investigated which NCT-CRC-HE-100 K dataset Kather_texture_2016_image_tiles dataset. The highest achieved accuracy reached 99.3% 96.8% ANOVA have successfully lowered dimensionality thus reducing complexity. Color-CADx has demonstrated efficacy terms accuracy, as its performance surpasses recent advancements.

Language: Английский

---

Ensemble Deep Learning-Based Image Classification for Breast Cancer Subtype and Invasiveness Diagnosis from Whole Slide Image Histopathology

Cancers, Journal Year: 2024, Volume and Issue: 16(12), P. 2222 - 2222

Published: June 14, 2024

Cancer diagnosis and classification are pivotal for effective patient management treatment planning. In this study, a comprehensive approach is presented utilizing ensemble deep learning techniques to analyze breast cancer histopathology images. Our datasets were based on two widely employed from different centers tasks: BACH BreakHis. Within the dataset, proposed strategy was employed, incorporating VGG16 ResNet50 architectures achieve precise of Introducing novel image patching technique preprocess high-resolution facilitated focused analysis localized regions interest. The annotated dataset encompassed 400 WSIs across four distinct classes: Normal, Benign, Situ Carcinoma, Invasive Carcinoma. addition, used BreakHis VGG16, ResNet34, models classify microscopic images into eight categories (four benign malignant). For both datasets, five-fold cross-validation rigorous training testing. Preliminary experimental results indicated patch accuracy 95.31% (for dataset) WSI 98.43% (BreakHis). This research significantly contributes ongoing endeavors in harnessing artificial intelligence advance diagnosis, potentially fostering improved outcomes alleviating healthcare burdens.

Language: Английский

---

Applying Explainable Machine Learning Models for Detection of Breast Cancer Lymph Node Metastasis in Patients Eligible for Neoadjuvant Treatment

Cancers, Journal Year: 2023, Volume and Issue: 15(3), P. 634 - 634

Published: Jan. 19, 2023

Background: Due to recent changes in breast cancer treatment strategy, significantly more patients are treated with neoadjuvant systemic therapy (NST). Radiological methods do not precisely determine axillary lymph node status, up 30% of being misdiagnosed. Hence, supplementary for status assessment needed. This study aimed apply and evaluate machine learning models on clinicopathological data, a focus meeting NST criteria, metastasis prediction. Methods: From the total patient data (n = 8381), 719 were identified as eligible NST. Machine applied NST-criteria group population. Model explainability was obtained by calculating Shapley values. Results: In group, random forest achieved highest performance (AUC: 0.793 [0.713, 0.865]), while population, XGBoost performed best 0.762 [0.726, 0.795]). values tumor size, Ki-67, age most important predictors. Conclusion: Tree-based achieve good assessing status. Such can lead accurate disease stage prediction consecutively better selection, especially where radiological clinical findings often only way assessment.

Language: Английский

---

Development of an Artificial Intelligence-Based Breast Cancer Detection Model by Combining Mammograms and Medical Health Records

Diagnostics, Journal Year: 2023, Volume and Issue: 13(3), P. 346 - 346

Published: Jan. 17, 2023

Background: Artificial intelligence (AI)-based computational models that analyze breast cancer have been developed for decades. The present study was implemented to investigate the accuracy and efficiency of combined mammography images clinical records detection using machine learning deep classifiers. Methods: This verified 731 from 357 women who underwent at least one mammogram had six months before mammography. model trained on mammograms variables discriminate benign malignant lesions. Multiple pre-trained CNN detect in mammograms, including X-ception, VGG16, ResNet-v2, ResNet50, CNN3 were employed. Machine constructed k-nearest neighbor (KNN), support vector (SVM), random forest (RF), Neural Network (ANN), gradient boosting (GBM) dataset. Results: performance obtained an 84.5% with a specificity 78.1% sensitivity 89.7% AUC 0.88. When image data alone, result achieved slightly lower score than (accuracy, 72.5% vs. 84.5%, respectively). Conclusions: A cancer-detection combining performed this satisfactory result, has potential applications.

Language: Английский

---

Extended Deep-Learning Network for Histopathological Image-Based Multiclass Breast Cancer Classification Using Residual Features

Symmetry, Journal Year: 2024, Volume and Issue: 16(5), P. 507 - 507

Published: April 23, 2024

Autonomy of breast cancer classification is a challenging problem, and early diagnosis highly important. Histopathology images provide microscopic-level details tissue samples play crucial role in the accurate cancer. Moreover, advancements deep learning an essential diagnosis. However, existing techniques involve unique models for each based on magnification factor require training numerous or using hierarchical approach combining multiple irrespective focus cell features. This may lead to lower performance multiclass categorization. paper adopts DenseNet161 network by adding learnable residual layer. The layer enhances features, providing low-level information. In addition, features are obtained from convolution preceding layer, which ensures that future size consistent with number channels DenseNet’s concatenation spatial helps better learn texture without need additional feature extraction module. model was validated both binary categorization malignant images. proposed model’s accuracy ranges 94.65% 100% classification, error rate 2.78%. Overall, suggested has potential improve survival patients allowing precise therapy.

Language: Английский

---

Advancements in Data Augmentation and Transfer Learning: A Comprehensive Survey to Address Data Scarcity Challenges

Recent Advances in Computer Science and Communications, Journal Year: 2024, Volume and Issue: 17(8)

Published: Jan. 11, 2024

Abstract: Deep Learning (DL) models have demonstrated remarkable proficiency in image classification and recognition tasks, surpassing human capabilities. The observed enhancement performance can be attributed to the utilization of extensive datasets. Nevertheless, DL huge data requirements. Widening learning capability such from limited samples even today remains a challenge, given intrinsic constraints small trifecta challenges, encompassing labeled datasets, privacy, poor generalization performance, costliness annotations, further compounds difficulty achieving robust model performance. Overcoming challenge expanding capabilities with sample sizes pressing concern today. To address this critical issue, our study conducts meticulous examination established methodologies, as Data Augmentation Transfer Learning, which offer promising solutions scarcity dilemmas. Augmentation, powerful technique, amplifies size datasets through diverse array strategies. These encompass geometric transformations, kernel filter manipulations, neural style transfer amalgamation, random erasing, Generative Adversarial Networks, augmentations feature space, adversarial meta- training paradigms. : Furthermore, emerges crucial tool, leveraging pre-trained facilitate knowledge between or enabling retraining on analogous Through comprehensive investigation, we provide profound insights into how synergistic application these two techniques significantly enhance effectively magnifying scarce This augmentation availability not only addresses immediate challenges posed by but also unlocks full potential working Big new era possibilities applications.

Language: Английский

---

Deep Crowd Anomaly Detection by Fusing Reconstruction and Prediction Networks

Electronics, Journal Year: 2023, Volume and Issue: 12(7), P. 1517 - 1517

Published: March 23, 2023

Abnormal event detection is one of the most challenging tasks in computer vision. Many existing deep anomaly models are based on reconstruction errors, where training phase performed using only videos normal events and model then capable to estimate frame-level scores for an unknown input. It assumed that error gap between frames abnormal high during testing phase. Yet, this assumption may not always hold due superior capacity generalization neural networks. In paper, we design a generalized framework (rpNet) proposing series by fusing several options network (rNet) prediction (pNet) detect efficiently. rNet, either convolutional autoencoder (ConvAE) or skip connected ConvAE (AEc) can be used, whereas pNet, traditional U-Net, non-local block attention U-Net (aUnet) applied. The fusion both rNet pNet increases gap. Our have distinct degree feature extraction capabilities. One our (AEcaUnet) consists AEc with proposed aUnet has capability confirm better extract quality features needed video detection. Experimental results UCSD-Ped1, UCSD-Ped2, CUHK-Avenue, ShanghaiTech-Campus, UMN datasets rigorous statistical analysis show effectiveness models.

Language: Английский

---

Analyzing Histological Images Using Hybrid Techniques for Early Detection of Multi-Class Breast Cancer Based on Fusion Features of CNN and Handcrafted

Diagnostics, Journal Year: 2023, Volume and Issue: 13(10), P. 1753 - 1753

Published: May 17, 2023

Breast cancer is the second most common type of among women, and it can threaten women's lives if not diagnosed early. There are many methods for detecting breast cancer, but they cannot distinguish between benign malignant tumors. Therefore, a biopsy taken from patient's abnormal tissue an effective way to challenges facing pathologists experts in diagnosing including addition some medical fluids various colors, direction sample, small number doctors their differing opinions. Thus, artificial intelligence techniques solve these help clinicians resolve diagnostic differences. In this study, three techniques, each with systems, were developed diagnose multi binary classes datasets types 40× 400× factors. The first technique dataset using neural network (ANN) selected features VGG-19 ResNet-18. by ANN combined ResNet-18 before after principal component analysis (PCA). third analyzing hybrid features. handcrafted; handcrafted. handcrafted mixed extracted Fuzzy color histogram (FCH), local pattern (LBP), discrete wavelet transform (DWT) gray level co-occurrence matrix (GLCM) methods. With data set, reached precision 95.86%, accuracy 97.3%, sensitivity 96.75%, AUC 99.37%, specificity 99.81% images at magnification factor 400×. Whereas 99.74%, 99.7%, 100%, 99.85%, 100%

Language: Английский

---

Fusing global context with multiscale context for enhanced breast cancer classification

Scientific Reports, Journal Year: 2024, Volume and Issue: 14(1)

Published: Nov. 9, 2024

Breast cancer is the second most common type of among women. Prompt detection breast can impede its advancement to more advanced phases, thereby elevating probability favorable treatment consequences. Histopathological images are commonly used for classification due their detailed cellular information. Existing diagnostic approaches rely on Convolutional Neural Networks (CNNs) which limited local context resulting in a lower accuracy. Therefore, we present fusion model composed Vision Transformer (ViT) and custom Atrous Spatial Pyramid Pooling (ASPP) network with an attention mechanism effectively classifying from histopathological images. ViT enables attain global features, while ASPP accommodates multiscale features. Fusing features derived models resulted robust classifier. With help five-stage image preprocessing technique, proposed achieved 100% accuracy BreakHis dataset at 100X 400X magnification factors. On 40X 200X magnifications, 99.25% 98.26% respectively. commendable efficacy images, be considered dependable option proficient classification.

Language: Английский

---