In the field of medical image recognition, the feature extraction of images is closely related to the magnification of the image, so most models of breast cancer image recognition will perform experiments at different magnifications. However, in practical applications, it is hoped that different magnifications of image information can be comprehensively utilized to comprehensively evaluate disease features and improve patient treatment effectiveness. In response to the above issues and the challenges of tumor classification in medical images, a classification model based on convolutional neural networks (CNN) and context-aware attentional pooling (CAP) is proposed, focusing on tumor categories without relying on specific magnifications. Firstly, the convolutional features of the image are extracted through CNN, and then the four levels of feature context information (including pixel-level, small-region, large-region and image-level) are comprehensively considered by combining them with the CAP module for classification. Using DenseNet121, MobileNetV2 and Xception three CNN networks combined with CAP, experiments were carried out the on BreakHis dataset. Four data of the same category with different magnifications were combined to identify eight types of breast cancer images. The accuracy of the model reached 96.87%, verifying its effectiveness in medical image classification.
SUNK, LEIL, ZHENGR, et al. Trends in incidence rates, mortality rates, and age-period-cohort effects of female breast cancer-China, 2003-2017[J]. China CDC Weekly, 2023, 5(15): 340-346.
[2]
SPANHOLF A, OLIVEIRAL S, PETITJEANC, et al. A dataset for breast cancer histopathological image classification[J]. IEEE Transactions on Biomedical Engineering, 2016, 63(7): 1455-1462.
[3]
HEK, ZHANGX, RENS, et al. Deep residual learning for image recognition[C]//2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2016: 770-778.
[4]
KONÉI, BOULMANEL. Hierarchical ResNeXt models for breast cancer histology image classification[M]//Lecture Notes in Computer Science. Cham: Springer International Publishing, 2018: 796-803.
[5]
BRANCATIN, FRUCCIM, RICCIOD. Multi-classification of breast cancer histology images by using a fine-tuning strategy[C]//International Conference Image Analysis and Recogition. Cham: Springer, 2018: 771-778.
[6]
RAKHLINA, SHVETSA, IGLOVIKOVV, et al. Deep convolutional neural networks for breast cancer histology image analysis[C]//International Conference Image Analysis and Recognition. Cham: Springer, 2018: 737-744.
[7]
FERREIRAC A, MELOT, SOUSAP, et al. Classification of breast cancer histology images through transfer learning using a pre-trained inception resnet V2[M]//Lecture Notes in Computer Science. Cham: Springer, 2018: 763-770.
LIANGMeiyan, ZHANGQiannan, RENZhuyun, et al. Research on identification of colon pathology image based on attention mechanism[J]. Journal of Test and Measurement Technology, 2022, 36(2): 93-100. (in Chinese)
[10]
BEHERAA, WHARTONZ, HEWAGEP R P G, et al. Context-aware attentional pooling (CAP) for fine-grained visual classification[C]//Proceedings of the AAAI Conference on Artificial Intelligence, 2021, 35(2): 929-937.
[11]
HUANGG, LIUZ, VAN DER MAATENL, et al. Densely connected convolutional networks[C]//2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2017: 2261-2269.
[12]
SANDLERM, HOWARDA, ZHUM, et al. MobileNetV2: inverted residuals and linear bottlenecks[C]//2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2018: 4510-4520.
[13]
CHOLLETF. Xception: deep learning with depthwise separable convolutions [C]//2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2017: 1800-1807.
[14]
HOCHREITERS, SCHMIDHUBERJ. Long short-term memory[J]. Neural Computation, 1997, 9: 1735-1780.
[15]
SELVARAJUR R, COGSWELLM, DAS A, et al. Grad-CAM: visual explanations from deep networks via gradient-based localization[J]. International Journal of Computer Vision (ICCV), 2020, 128(2): 336-359.
[16]
KRIZHEVSKYA, SUTSKEVERI, HINTONG E. ImageNet classification with deep convolutional neural networks[J]. Communications of the ACM, 2017, 60(6): 84-90.
[17]
SIMONYANK, ZISSERMANA. Very deep convolutional networks for large-scale image recognition[J/OL]. Computer Science, 2014.DOI:10.48550/arXiv.1409.1556 .
[18]
HEK, ZHANGX, RENS, et al. Deep residual learning for image recognition [C]//2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2016: 770-778.
[19]
SZEGEDYC, LIUW, JIAY, et al. Going deeper with convolutions [C]//2015 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2015: 1-9.
[20]
SZEGEDYC, VANHOUCKEV, IOFFES, et al. Rethinking the inception architecture for computer vision [C]//2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2016: 2818-2826.
[21]
BENHAMMOUY, ACHCHABB, HERRERAF, et al. BreakHis based breast cancer automatic diagnosis using deep learning: taxonomy, survey and insights[J]. Neurocomputing, 2020, 375: 9-24.
[22]
LIUW, JUHASM, ZHANGY. Fine-grained breast cancer classification with bilinear convolutional neural networks (BCNNs)[J]. Frontiers in Genetics, 2020, 11: 547327.
[23]
BOUMARAFS, LIUX, ZHENGZ, et al. A new transfer learning based approach to magnification dependent and independent classification of breast cancer in histopathological images [J]. Biomedical Signal Processing and Control, 2021, 63: 102192.
ZHAOXiaoping, WANGRongfa, SUNZhongbo, et al. Research on eight classifications of breast cancer pathological images based on improved DenseNet [J]. Computer Engineering and Applications, 2023, 59(5): 213-221. (in Chinese)