Research On Assisted Diagnostic Method For Stones In Body Based On Image Group

In recent years,the incidence of urinary tract stones in China has been increasing year by year.In terms of simple stones in the body,calcium oxalate is the most common,followed by anhydrous uric acid.At present,traditional methods require the instruments to distinguish them and the steps are complicated and costly.At the same time,non-invasive detection in vivo before surgery is impossibleAt present,the assisted diagnosis method based on imaging omics has developed from a pure theory to a clinical trial,but its research on the identification of stone components is less.Firstly,in the pre-processing stage,when the traditional sequence interpolation algorithm performs three-dimensional reconstruction of the lesion,the effect is poor and the calculation is complicated,and it cannot take into account the grayscale and shape changes of the lesion image.Secondly,in the segmentation stage,because medical images are more detailed and complex than traditional images,and their segmentation methods are still based on supervised or semi-supervised learning,traditional segmentation algorithms are not universally used for stone lesions.Moreover,the algorithm in the follow-up stage fails to fully consider the clinical characteristics of the calculus in vivo,and the identification effect is poor.In view of the above,the research work of this paper is as follows(1)An improved sequence interpolation algorithm based on wavelet subgraphs is proposed:the algorithm first adjusts the grayscale of computed tomography(CT)images of stones,and then selects the appropriate mother wavelet and wavelet transform function to decompose the original CT image into low-frequency sub-pictures and high-frequency sub-pictures,low-frequency sub-pictures are interpolated using a bilinear interpolation algorithm to smooth the image so as to highlight the large areas,and the original bilinear interpolation algorithm is improved for high-frequency sub-pictures to highlight edge details in high-frequency sub-pictures.Finally,the wavelet sub-graphs after interpolation are reconstructed to generate new interpolation images for better 3D reconstruction.(2)An improved image segmentation algorithm based on the Unet network is proposed:before performing network model training,we firstly perform grayscale adjustment and normalization on the CT image,and then perform data enhancement through translation,rotation,and zoom operations to meet the massive training set required by neural network during training.Then,the traditional Unet network is improved,that is,the components of the network are optimized,and appropriate training strategies are formulated to obtain a segmentation model that meets the characteristics of stones so as to automatically segment the region of interest of the lesion(3)An improved assisted diagnostic method for stone composition in vivo based on imaging omics is proposed:according to the clinical characteristics of stone lesions,on the basis of improving the pre-processing and segmentation algorithms,some suitable algorithms are selected for imaging omics improvements in subsequent phases.First,the pre-processed and segmented stone lesions are extracted in two-dimensional and three-dimensional gray,shape,and texture features respectively,and the feature information in each field is complementary to comprehensively improve the method's discrimination capability.Then,it is screened by a feature selection algorithm,and put the selected features into the classifier for training,and finally get the auxiliary diagnosis method of the stones in the body.Through the experimental comparative analysis,the effectiveness and feasibility of the method for distinguishing calcium oxalate and anhydrous uric acid stones in vivo were verified.Experiments and results show that the accuracy(ACC)and the area under curve(AUC)values of the final auxiliary diagnostic method reach 81.76%and 89.03%respectively,which can effectively analyze the simple calcium oxalate and anhydrous uric acid components in the stones in the body,and provide an effective reference for the diagnosis of clinicians.