Research And Implementation Of CAP System For Axillary Lymph Node Metastasis In Early Breast Cancer

Breast cancer is one of the most common malignant tumors in women.Due to cancer cells are easily shed and migrate to normal tissues and organs,such as lymph nodes.Preoperative prediction of axillary lymph node status in patients with early breast cancer plays great significant role in selecting the best treatment.The main Axillary Lymph Node metastasis screening and treatment options are invasive Sentinel Lymph Node activity detection(SLNB)and Axillary Lymph Node Dissection(ALND).These measurements usually lead to upper arm edema,upper limb pain and other adverse reactions.Ultrasound has the characteristics of noninvasive and radiation,and it can avoid the above adverse reactions.This paper studies and implements a Computer Aided Prediction(CAP)system for axillary lymph node metastasis of early breast cancer based on breast ultrasound images,which can assist in predicting axillary lymph node metastasis of early breast cancer patients and improve the survival rate of early breast cancer patients.The main research work of this article is as follows:(1)Designed and realized an automatic image segmentation algorithm Pyramid Pooling U-Net(PPU-Net).In the PPU-Net,the pyramid pooling module was applied together with the U-Net to extract more scale image information.The experiment proved that the algorithm was feasible in the breast gray ultrasound image segmentation and improved the accuracy of image segmentation.(2)Realized the elastic ultrasound image reconstruction,the Red-Green-Blue(RGB)value of 3d color elastic image is reconstructed into one dimensional elastic modulus.Designed the characteristic flow of mammary gland image to extract morphological,grayscale,texture,calcification and elastic features.(3)Designed and implemented a feature selection algorithm based on multi-angle fusion strategy.In this algorithm,Least Absolute Shrinkage and Selection Operator,LASSO(LASSO)feature selection algorithm,Mutual Information(MI)feature selection algorithm and random forest feature selection algorithm are fused to built a strong robustness and high accuracy feature selection model.The fused feature selection algorithm can make up for the shortcomings of the above three algorithms.The experiment proved that this algorithm was feasible in the axillary lymph node metastasis prediction of the early breast cancer.It also improved the accuracy and sensitivity of the prediction results based on Support Vector Machine(SVM)classifier.(4)Designed and implemented the CAP system for axillary lymph node metastasis of early breast cancer.On the basis of the overall scheme design of the system and the design of each functional module,the probabilistic analysis of metastasis prediction and the statistical analysis of important features were analyzed.