The efficacy of combining Deep and Handcrafted Features for Breast Cancer Classification using Ultrasound Images DOI
Barsha Abhisheka, Saroj Kr. Biswas, Debasmita Saha

и другие.

2018 5th IEEE Uttar Pradesh Section International Conference on Electrical, Electronics and Computer Engineering (UPCON), Год журнала: 2023, Номер unknown, С. 735 - 740

Опубликована: Дек. 1, 2023

Breast cancer is a potentially fatal condition, and timely detection plays vital role in enhancing survival rates. To address this issue aid clinicians early detection, Computer Aided systems have been explored. Many researchers proposed solutions for breast popular approach involves using Convolutional Neural Networks (CNNs). CNN-based approaches displayed encouraging outcomes owing to their capacity autonomously capture advanced features from medical images. But relying solely on global might lead suboptimal classification results, as local image details may be overlooked. improve the performance, paper introduces system called Multi Featured Cancer Detection System (MFBCDS). This takes advantage of both CNN handcrafted features. Histogram Oriented Gradient (HOG), Local Binary Pattern (LBP) are utilized extracting features, while obtained ResNet50. The MFBCDS model integrates create comprehensive feature vector. vector captures essential information localized regions, complementing extracted by CNN. Therefore, combination significantly enhances performance model. combined classified Support Vector Machine (SVM). evaluation carried out widely used BUSI dataset 5-fold cross-validation where has achieved satisfactory various matrices with an average accuracy 88.87%.

Язык: Английский

AI in Breast Cancer Imaging: An Update and Future Trends DOI Creative Commons
Yizhou Chen, Xiaoliang Shao, Kuangyu Shi

и другие.

Seminars in Nuclear Medicine, Год журнала: 2025, Номер unknown

Опубликована: Фев. 1, 2025

Язык: Английский

Процитировано

2
Evaluating the Role of Breast Ultrasound in Early Detection of Breast Cancer in Low- and Middle-Income Countries: A Comprehensive Narrative Review DOI Creative Commons
Roxana Iacob, Emil Radu Iacob, Emil Robert Stoicescu

и другие.

Bioengineering, Год журнала: 2024, Номер 11(3), С. 262 - 262

Опубликована: Март 7, 2024

Breast cancer, affecting both genders, but mostly females, exhibits shifting demographic patterns, with an increasing incidence in younger age groups. Early identification through mammography, clinical examinations, and breast self-exams enhances treatment efficacy, challenges persist low- medium-income countries due to limited imaging resources. This review assesses the feasibility of employing ultrasound as primary cancer screening method, particularly resource-constrained regions. Following PRISMA guidelines, this study examines 52 publications from last five years. ultrasound, distinct offers advantages like radiation-free imaging, suitability for repeated screenings, preference populations. Real-time dense tissue evaluation enhance sensitivity, accessibility, cost-effectiveness. However, limitations include reduced specificity, operator dependence, detecting microcalcifications. Automatic (ABUS) addresses some issues faces constraints potential inaccuracies microcalcification detection. The analysis underscores need a comprehensive approach screening, emphasizing international collaboration addressing limitations, especially settings. Despite advancements, notably ABUS, goal is contribute insights optimizing globally, improving outcomes, mitigating impact debilitating disease.

Язык: Английский

Процитировано

11
BraNet: a mobil application for breast image classification based on deep learning algorithms DOI
Yuliana Jiménez-Gaona,

María José Rodríguez Álvarez,

Darwin Castillo

и другие.

Medical & Biological Engineering & Computing, Год журнала: 2024, Номер 62(9), С. 2737 - 2756

Опубликована: Май 2, 2024

Язык: Английский

Процитировано

6
Enhancing the fairness of AI prediction models by Quasi-Pareto improvement among heterogeneous thyroid nodule population DOI Creative Commons
Siqiong Yao, Fang Dai, Peng Sun

и другие.

Nature Communications, Год журнала: 2024, Номер 15(1)

Опубликована: Март 4, 2024

Abstract Artificial Intelligence (AI) models for medical diagnosis often face challenges of generalizability and fairness. We highlighted the algorithmic unfairness in a large thyroid ultrasound dataset with significant diagnostic performance disparities across subgroups linked causally to sample size imbalances. To address this, we introduced Quasi-Pareto Improvement (QPI) approach deep learning implementation (QP-Net) combining multi-task domain adaptation improve model among disadvantaged without compromising overall population performance. On dataset, our method significantly mitigated area under curve (AUC) disparity three less-prevalent by 0.213, 0.112, 0.173 while maintaining AUC dominant subgroups; also further confirmed on two public datasets: ISIC2019 skin disease CheXpert chest radiograph dataset. Here show QPI be widely applicable promoting AI equitable healthcare outcomes.

Язык: Английский

Процитировано

4
Analysis of the diagnostic efficacy of ultrasound, MRI, and combined examination in benign and malignant breast tumors DOI Creative Commons
Di Ma, Changliang Wang, Jie Li

и другие.

Frontiers in Oncology, Год журнала: 2025, Номер 15

Опубликована: Янв. 30, 2025

Background To compare the diagnostic effectiveness of ultrasound (US), magnetic resonance imaging (MRI), and their combined application in distinguishing between benign malignant breast tumors, with particular emphasis on evaluating performance different densities—fatty tissue, where fat predominates, dense which contains a significant amount fibroglandular tissue. Materials methods A retrospective analysis was conducted 185 patients including 90 95 cases. All underwent both US MRI examinations within one week prior to surgery. The accuracy US, MRI, use differentiating tumors evaluated. Results examination demonstrated highest area under curve (AUC), sensitivity, negative predictive value (NPV) (0.904, 90%, 90.4%), outperforming (0.830, 73.3%, 78.6%) (0.897, 89.7%, 88.8%). DeLong test results revealed statistically differences AUC as well (P < 0.05). However, difference not = 0.939). In fatty no were found or 0.708 P 0.317, respectively). For examination, 0.05), while 0.317). Conclusion significantly enhance ability differentiate provide important clinical for early cancer detection.

Язык: Английский

Процитировано

0
Diagnosis of Benign and Malignant Newly Developed Nodules on the Surgical Side After Breast Cancer Surgery Based on Machine Learning DOI Creative Commons
Zhixiang Wang, Qingqing Li, Yiran Wang

и другие.

The Breast Journal, Год журнала: 2025, Номер 2025(1)

Опубликована: Янв. 1, 2025

Objective: To enhance the diagnostic accuracy of new nodules on surgical side after breast cancer surgery using machine learning techniques and to explore role multifeature fusion. Methods: Data from 137 postoperative patients with January 2016 April 2024 were analyzed. Clinical, ultrasound, immunohistochemistry, features combined. Multiple models, including support vector (SVM), random forest, gradient boosting, AdaBoost, XGBoost, trained tested. Model performance was evaluated stratified ten-fold cross-validation. Ablation experiments assessed impact different feature combinations performance. Results: The SVM model performed best, an AUC 0.8664, 0.8099, a sensitivity 0.565, specificity 0.9267. indicated that fusion significantly improved performance, especially when combining clinical, features. Gradient boosting forest models showed slightly inferior while AdaBoost had balanced but lower effectiveness. Conclusion: Machine learning, particularly model, shows significant potential in diagnosing surgery. It can assist doctors developing more effective treatment plans, improving patient outcomes. Future studies should expand sample sizes, include multicenter data, advanced algorithms further

Язык: Английский

Процитировано

0
A deep learning-based multimodal medical imaging model for breast cancer screening DOI Creative Commons

Junwei Chen,

Teng Pan,

Zhengjie Zhu

и другие.

Scientific Reports, Год журнала: 2025, Номер 15(1)

Опубликована: Апрель 26, 2025

In existing breast cancer prediction research, most models rely solely on a single type of imaging data, which limits their performance. To overcome this limitation, the present study explores based multimodal medical images (mammography and ultrasound images) compares them with single-modal models. We collected data from 790 patients, including 2,235 mammography 1,348 images, conducted comparison using six deep learning classification to identify best model for constructing model. Performance was evaluated metrics such as area under receiver operating characteristic curve (AUC), sensitivity, specificity, precision, accuracy compare Experimental results demonstrate that outperforms in terms specificity (96.41% (95% CI:93.10%-99.72%)), (93.78% CI:87.67%-99.89%)), precision (83.66% CI:76.27%-91.05%)), AUC (0.968 CI:0.947-0.989)), while excel sensitivity. Additionally, heatmap visualization used further validate performance conclusion, our shows strong potential screening tasks, effectively assisting physicians improving accuracy.

Язык: Английский

Процитировано

0
A survey on deep learning in medical ultrasound imaging DOI Creative Commons
Ke Song, Jing Feng, Duo Chen

и другие.

Frontiers in Physics, Год журнала: 2024, Номер 12

Опубликована: Июль 1, 2024

Ultrasound imaging has a history of several decades. With its non-invasive, low-cost advantages, this technology been widely used in medicine and there have many significant breakthroughs ultrasound imaging. Even so, are still some drawbacks. Therefore, novel image reconstruction analysis algorithms proposed to solve these problems. Although new solutions effects, them introduce other side such as high computational complexity beamforming. At the same time, usage requirements medical equipment relatively high, it is not very user-friendly for inexperienced beginners. As artificial intelligence advances, researchers initiated efforts deploy deep learning address challenges imaging, reducing adaptive beamforming aiding novices acquisition. In survey, we about explore application spanning from clinical diagnosis.

Язык: Английский

Процитировано

3
TDF-Net: Trusted Dynamic Feature Fusion Network for breast cancer diagnosis using incomplete multimodal ultrasound DOI
Pengfei Yan,

Wushuang Gong,

Minglei Li

и другие.

Information Fusion, Год журнала: 2024, Номер 112, С. 102592 - 102592

Опубликована: Июль 20, 2024

Язык: Английский

Процитировано

3
Gray-to-color image conversion in the classification of breast lesions on ultrasound using pre-trained deep neural networks DOI
Wilfrido Gómez‐Flores, Wagner Coelho de Albuquerque Pereira

Medical & Biological Engineering & Computing, Год журнала: 2023, Номер 61(12), С. 3193 - 3207

Опубликована: Сен. 15, 2023

Язык: Английский

Процитировано

5