Pulmonary CT Image Filtering And Pulmonary Nodules Segmentation Performance Verification Of Filtering Effect

Lung cancer is the most common tumor in the world today.Detection and segmentation of lung nodules is the key technology for early detection and diagnosis of lung tumors.Computed Tomography(CT)is one of the most widely used medical imaging methods in the diagnosis of lung cancer.However,in the process of CT examination,due to the influence of various factors such as external environment,the CT image inevitably introduces a lot of noise in the imaging process.The direct use of unfiltered CT images for analysis will cause great interference to doctors’ diagnosis.In addition,the subsequent processing of CT images may also cause unnecessary errors due to the superposition of noise.The segmentation results of pulmonary nodules depend very much on the quality and signal-to-noise ratio of lung CT images,so it is very necessary to filter CT images to improve the image quality.The purpose of this paper is to study an image filtering method with denoising and good retention of image edge information in the stage of CT image preprocessing.The main research contents are as follows:(1)The median filter and its improved filter were used to filter the lung CT images.By changing the values of all pixels in the set filtering window,the value of the same pixel can be represented by adding variable information greater than one,and then the overall median value of the expanded digital set can be obtained.This kind of filter for filtering the noise of the image point only,not to get rid of the image signal,the image can be secondary screening,to avoid the edge information point being mistaken for noise cancelling,to a certain extent,prevent the edge information,the filter will be lost in the after image denoising images to obtain a better effect.(2)Combined with the effect of Laplace algorithm on image enhancement,the original image and Laplace image are added together to achieve the effect of sharpening the image,and finally the details are highlighted on the premise of retaining the background information of the image.According to the shape of a gaussian function option value,and finally to smooth the image noise,make the image of the gray level is worth to retain,making gray mutation of contrast enhanced,on the premise of image background,highlights the image of the small and medium-sized details,excellent in denoising for lung CT image at the same time retain the image edge information.(3)The image quality after filtering is evaluated by four indicators.The lung sample images were verified by MATLAB software.We also used U-Net Convolutional Neural Network to segment lung nodules from images preprocessed by different filters,and verified the performance of the filters by analyzing the subsequent segmentation effects of CT images preprocessed by different filters.The results show that the Gaussian Laplace filter has the best performance.Compared with the processing effect of other filters,the Gaussian Laplace filter can better retain the edge information of the image while denoising,and has the best effect in the subsequent segmentation processing,so it is the most suitable for medical CT images.