Deep learning-based feature extraction and discrimination of benign and malignant tumors in breast ultrasound images
10.3969/j.issn.1005-202X.2025.10.011
- VernacularTitle:乳腺超声图像的深度学习特征提取与良恶性肿瘤鉴别
- Author:
Ni'na DAI
1
;
Wenjun ZHANG
1
Author Information
1. 湖北医药学院附属太和医院超声医学科,湖北 十堰 442000
- Publication Type:Journal Article
- Keywords:
breast ultrasound image;
convolutional neural network;
transfer learning;
feature extraction
- From:
Chinese Journal of Medical Physics
2025;42(10):1342-1347
- CountryChina
- Language:Chinese
-
Abstract:
Considering the complexity and diversity of breast ultrasound images,a deep learning model combining convolutional neural network and transfer learning is proposed to enhance the accuracy and efficiency of tumor discrimination.Specifically,a pre-trained convolutional neural network model is utilized to extract features from breast ultrasound images,and transfer learning is employed to adapt the pre-trained model to the breast ultrasound image dataset.The extracted features are then applied to multiple classification models to differentiate between benign and malignant tumors.Experimental results demonstrate that compared with traditional image processing and machine learning methods,the proposed deep learning model exhibits significant improvements in sensitivity,specificity,and precision.The deep learning-based feature extraction method for breast ultrasound images substantially enhances the accuracy and efficiency of discrimination between benign and malignant tumors,providing an effective technical tool for the early diagnosis and treatment of breast cancer and assisting clinicians in making more accurate clinical decisions.
