Feature Fusion Based on Convolutional Neural Network for Breast Cancer Auxiliary Diagnosis

Abstract

Cancer is one of the leading causes of death in many countries. Breast cancer is one of the most common cancers in women. Especially in remote areas with low medical standards, the diagnosis efficiency of breast cancer is extremely low due to insufficient medical facilities and doctors. Therefore, in-depth research on how to improve the diagnosis rate of breast cancer has become a hot spot. With the development of society and science, people use artificial intelligence to improve the auxiliary diagnosis of diseases in the existing medical system, which can become a solution for detecting and accurately diagnosing breast cancer. The paper proposes an auxiliary diagnosis model that uses deep learning in view of the low rate of human diagnosis by doctors in remote areas. The model uses classic convolutional neural networks, including VGG16, InceptionV3, and ResNet50 to extract breast cancer image features, then merge these features, and finally train the model VIRNets for auxiliary diagnosis. Experimental results prove that for the recognition of benign and malignant breast cancer pathological images under different magnifications, VIRNets have a high generalization and strong robustness, and their accuracy is better than their basic network and other structures of the network. Therefore, the solution provides a certain practical value for assisting doctors in the diagnosis of breast cancer in real scenes.

Efficient breast cancer diagnosis using multi‐level progressive feature aggregation based deep transfer learning system

“…Feature fusion is a robust and efficient approach to improve the classification performance by reducing the problem of vanishing gradient through retainment of intermediate feature maps. Cheng et al 50 have introduced a feature fusion‐based CNN for BC diagnosis, which is a multi‐network model and reporting accuracy of 99.02% and 98.37% for 40× and 400× size of HIs, respectively. In, 51 a CNN‐based model is suggested for BC diagnosis from ultrasound images where the features from radiomics and CNN are fused to get better features.…”

“…Existing feature fusion methods of BC detection use TL‐based multinetwork feature fusion technique but the vanishing gradient issue still persists. It results compromising performance of the system even with high complexity 50–53 …”

Using Faster R-CNN to Detect and Recognize Arabic Handwritten Words

“…Therefore, the network's dimension should be changed. The input dimension of the identity block is the same as the output dimension, which may be connected in series and used to deepen the network [25,26]. Fig.…”

Hepatocellular Carcinoma Recognition from Ultrasound Images Using Combinations of Conventional and Deep Learning Techniques

“…Other approaches combined multiple types of deep learning features. In [23], aiming to detect breast cancer within histopathology images, the authors combined the deep learning features provided by the VGG16, Incep-tionV3 and ResNet50 architectures, the concatenated features being provided to a VirNet model, which performed the final feature fusion and classification. In the approach described in [24], the authors combined the deep learning features provided by the ResNet50 and DenseNet201 architectures for performing brain tumor classification.…”