Classification Of Breast Tumor Image Based On Fuzzy Logic And Deep Learning

Breast cancer,as one of the most common malignant tumors in women worldwide,lacks effective preventive measures so far,and can only rely on early detection and early diagnosis of patients to reduce mortality.In the field of deep learning,there are two main problems.First,the details of medical images will affect the accuracy of medical image processing results.Second,the deep learning method itself is a black box,which is unexplainable and lacks theoretical support.Because of the problems existing in the existing breast tumor diagnosis technology,the main research work and results of this article are as follows:(1)During image preprocessing,retain as much image detail information as possible under normalization,so that the image is in high-definition,fully consider that medical images should pay more attention to details and privacy than natural images so that the model is more suitable for medicine Image Processing.(2)Based on research and analysis of the breast tumor classification model based on deep learning,a breast tumor classification model and algorithm based on the multi-view convolutional neural network were constructed.Using a multi-view convolutional neural network breast tumor classification algorithm to train breast tumor images,the features obtained afer training are cross-fused to obtain more image information,and can effectively improve the accuracy of judgment of benign and malignant breast tumors.(3)Using fuzzy logic theory to improve deep learning models and algorithms,We created a breast cancer classification model and algorithm that can combine fuzzy logic theory and deep learning-an interpretable convolutional neural network method.(4)The interpretable convolutional neural network method is applied to breast tumor classification,and a breast tumor classification algorithm based on an interpretable convolutional neural network is proposed.Use fuzzy logic to explain the bias b in the convolutional neural network model,and use it to set the constraints with inputs and weights,reduce the amount of calculation,no longer need to learn optimization,and learn features faster.Experiments show that,compared with other algorithms,this algorithm has higher classification accuracy and better stability.While improving speed and accuracy,and promoted the interpretability of one of the difficult problems of deep learning in the medical field.