Research On Segmentation And Visualization Framework Of Lung Nodules Detection

Lung cancer is one of the most lethal cancers.Early detection and prevention can help to reduce the mortality of lung cancer patients.With the development of science and technology,a computer-aided detection system is gradually used in medicine to automatically segment the focus and assist doctors to analyze the disease,so as to improve the accuracy of doctors’ diagnosis.Although the computer-aided detection system improves the accuracy of doctors’ diagnosis,the irregular shape,size and location of pulmonary nodules bring difficulties to the detection system,in order to solve this problem,this paper proposes an improved U-Net algorithm for lung nodule segmentation,and carries out the three-dimensional visualization of lung nodules,the main contributions are as follows:(1)In order for the data set to be suitable for algorithmic input and to reduce the impact of other organizations,this paper Data pre-processing the LIDC-IDRI(Lung Image Database Consortium Image Database Resource Initiative).The Data pre-processing are mainly based on the order of two-dimensional slices,using threshold segmentation,some algorithms,such as morphology,process the data to obtain the training data in line with the algorithm in this paper.(2)Firstly,a RAU-Net(U-Net network with residual feature and attention mechanism)neural network model is proposed based on the improved U-Net algorithm,by adding residual elements and attention mechanism,the problem of gradient disappearance and low feature utilization of U-Net is solved,and the segment ation effect of U-Net on pulmonary nodules is improved.Finally,the experiment shows that the Meaniou of the RAU-Net neural network model designed in this paper can reach 91.13%,which greatly improves the segmentation accuracy of U-Net for pulmonary nodules.(3)In order to display the lung nodules more directly,a set of two-dimensional sections were transformed into three-dimensional display by using VTK tool kit,and the three-dimensional visualization of the lung nodules was realized.