Few-shot classification of ultrasound breast cancer images using meta-learning algorithms

Neural Computing and Applications, Journal Year: 2024, Volume and Issue: 36(20), P. 12047 - 12059

Published: April 18, 2024

Abstract Medical datasets often have a skewed class distribution and lack of high-quality annotated images. However, deep learning methods require large amount labeled data for classification. In this study, we present few-shot approach the classification ultrasound breast cancer images using meta-learning methods. We used prototypical networks model agnostic (MAML) algorithms as The (BUSI) dataset, which has three classes is difficult to use in meta-learning, was meta-testing cross-domain along with other meta-training. Our proposed yielded an accuracy range 0.882–0.889, achieved by implementing ResNet50 backbone ProtoNet 10-shot setting. These results represent significant improvement ranging from 6.27 7.10% over baseline 0.831. showed that outperformed MAML method all k-shot settings. addition, ResNet models network feature extraction found be more successful than four-layer convolutional model. first attempt apply BUSI dataset while providing higher compared medical small-scale few classes. methodology study can adapted similar problems.

Language: Английский

---

Intelligent Ultrasound Imaging for Enhanced Breast Cancer Diagnosis: Ensemble Transfer Learning Strategies

IEEE Access, Journal Year: 2024, Volume and Issue: 12, P. 22243 - 22263

Published: Jan. 1, 2024

According to WHO statistics for 2018, there are 1.2 million cases and 700,000 deaths from breast cancer (BC) each year, making it the second-highest cause of mortality women globally. In recent years, advances in artificial (AI) intelligence machine (ML) learning have shown incredible potential increasing accuracy efficiency BC diagnosis. This research describes an intelligent image analysis system that leverages capabilities transfer (TLs) with ensemble stacking ML models. As part this research, we created a model analyzing ultrasound images using cutting-edge TL models such as Inception V3, VGG-19, VGG-16. We implemented models, including MLP (Multi-Layer Perceptron) different architectures (10 10, 20 20, 30 30) Support Vector Machines (SVM) RBF Polynomial kernels. analyzed effectiveness proposed performance parameters (accuracy (CA), sensitivity, specificity, AUC). Compared results existing diagnostic systems, method (Inception V3 + Staking) is superior, 0.947 AUC 0.858 CA values. The BCUI consists data collection, pre-processing, learning, evaluation, comparative demonstrating its superiority over methods.

Language: Английский

---

Fine tuning deep learning models for breast tumor classification

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

Published: May 10, 2024

Abstract This paper proposes an approach to enhance the differentiation task between benign and malignant Breast Tumors (BT) using histopathology images from BreakHis dataset. The main stages involve preprocessing, which encompasses image resizing, data partitioning (training testing sets), followed by augmentation techniques. Both feature extraction classification tasks are employed a Custom CNN. experimental results show that proposed CNN model exhibits better performance with accuracy of 84% than applying same other pretrained models, including MobileNetV3, EfficientNetB0, Vgg16, ResNet50V2, present relatively lower accuracies, ranging 74 82%; these four models used as both extractors classifiers. To increase metrics, Grey Wolf Optimization (GWO), Modified Gorilla Troops (MGTO) metaheuristic optimizers applied each separately for hyperparameter tuning. In this case, model, refined MGTO optimization, reaches exceptional 93.13% in just 10 iterations, outperforming state-of-the-art methods, based on

Language: Английский

---

Revolutionizing breast ultrasound diagnostics with EfficientNet-B7 and Explainable AI

BMC Medical Imaging, Journal Year: 2024, Volume and Issue: 24(1)

Published: Sept. 2, 2024

Breast cancer is a leading cause of mortality among women globally, necessitating precise classification breast ultrasound images for early diagnosis and treatment. Traditional methods using CNN architectures such as VGG, ResNet, DenseNet, though somewhat effective, often struggle with class imbalances subtle texture variations, to reduced accuracy minority classes malignant tumors. To address these issues, we propose methodology that leverages EfficientNet-B7, scalable architecture, combined advanced data augmentation techniques enhance representation improve model robustness. Our approach involves fine-tuning EfficientNet-B7 on the BUSI dataset, implementing RandomHorizontalFlip, RandomRotation, ColorJitter balance dataset The training process includes stopping prevent overfitting optimize performance metrics. Additionally, integrate Explainable AI (XAI) techniques, Grad-CAM, interpretability transparency model's predictions, providing visual quantitative insights into features regions influencing outcomes. achieves 99.14%, significantly outperforming existing CNN-based approaches in image classification. incorporation XAI enhances our understanding decision-making process, thereby increasing its reliability facilitating clinical adoption. This comprehensive framework offers robust interpretable tool detection cancer, advancing capabilities automated diagnostic systems supporting processes.

Language: Английский

---

Symmetry in Privacy-Based Healthcare: A Review of Skin Cancer Detection and Classification Using Federated Learning

Symmetry, Journal Year: 2023, Volume and Issue: 15(7), P. 1369 - 1369

Published: July 5, 2023

Skin cancer represents one of the most lethal and prevalent types observed in human population. When diagnosed its early stages, melanoma, a form skin cancer, can be effectively treated cured. Machine learning algorithms play crucial role facilitating timely detection aiding accurate diagnosis appropriate treatment patients. However, implementation traditional machine approaches for disease is impeded by privacy regulations, which necessitate centralized processing patient data cloud environments. To overcome challenges associated with privacy, federated emerges as promising solution, enabling development privacy-aware healthcare systems diagnosis. This paper presents comprehensive review that examines obstacles faced conventional explores integration context privacy-conscious prediction systems. It provides discussion on various datasets available performance comparison techniques lesion prediction. The objective to highlight advantages offered potential addressing concerns realm

Language: Английский

---

Integration of ultrasound and mammogram for multimodal classification of breast cancer using hybrid residual neural network and machine learning

Image and Vision Computing, Journal Year: 2024, Volume and Issue: 145, P. 104987 - 104987

Published: March 12, 2024

Language: Английский

---

Effectiveness Analysis of Deep Learning Methods for Breast Cancer Diagnosis Based on Histopathology Images

Applied Sciences, Journal Year: 2025, Volume and Issue: 15(3), P. 1005 - 1005

Published: Jan. 21, 2025

The early detection of breast cancer is crucial for both accelerating the treatment process and preventing spread cancer. accuracy diagnosis also significantly influenced by experience pathologists. Many studies have been conducted on correct to help specialists increase diagnosis. This study focuses classifying using deep learning models, including pre-trained VGG16, MobileNet, DenseNet201, a custom-built Convolutional Neural Network (CNN), with final dense layer optimized via particle swarm optimization (PSO) algorithm. Breast Histopathology Images Dataset was used evaluate performance model, forming two datasets: one 157,572 images at 50 × 3 (Experimental Study 1) another 1116 resized 224 2). Both original (50 3) rescaled (224 were tested. highest success rate obtained CNN model an 93.80% experimental 1. MobileNet yielded 95.54% 2. results demonstrate that proposed exhibits promising, superior classification compared state-of-the-art methods across varying image sizes dataset volumes.

Language: Английский

---

Improving breast cancer classification in fine-grain ultrasound images through feature discrimination and a transfer learning approach

Biomedical Signal Processing and Control, Journal Year: 2025, Volume and Issue: 106, P. 107690 - 107690

Published: Feb. 20, 2025

Language: Английский

---

Ensemble Learning with Mixup Style for Ultrasound Image Classification

Communications in computer and information science, Journal Year: 2025, Volume and Issue: unknown, P. 261 - 271

Published: Jan. 1, 2025

Language: Английский

---

HAFMAB-Net: hierarchical adaptive fusion based on multilevel attention-enhanced bottleneck neural network for breast histopathological cancer classification

Signal Image and Video Processing, Journal Year: 2025, Volume and Issue: 19(5)

Published: March 19, 2025

Language: Английский

---