Volume Data Analysis And Modeling Of 3D Medical Image Based On Inter-Layer Feature Learning

Medical Image plays an essential role in medical research,clinical diagnosis and treatment.Medical imaging system imaging of the human body's internal tissues and organs,helps doctors diagnose and treat patients accurately.Since X-ray was discovered in1895,computerized tomography(CT)and magnetic resonance imaging(MRI)have been developed in medical imaging technology.Although the imaging mechanism of the above techniques is different,they all play essential roles in clinical diagnosis,treatment,and medical research.However,in the application of CT and MRI image,the traditional 2D display can only present 2D plane information,and doctors can not obtain more information about the lesion area from more angles.At the same time,limited by objective factors such as the performance of imaging equipment and the amount of medical does,the inter-layer resolution of medical image is far less than the intra-layer resolution,and the time cost of medical image scanning and reconstruction is high,which limits its application.If through the analysis of the change law of in layer plane features and inter layer features in 3D medical image volume data,the focus modeling can be realized by increasing the volume data,the diagnosis and treatment process can be more efficient,accurate and reliable.It also can improve the degree of information acquisition of scientific researchers in the process of research.The inter-layer resolution of medical images can be improved by inter-layer interpolation to analyze and model 3D medical image volume data effectively.This paper aims to enhance the 3D volume data of medical images utilizing the inter-layer interpolation technique.By increasing the number of images in the inter-layer dimension,the image inter-layer interpolation method improves the image inter-layer resolution.At present,there are two directions for image interpolation: traditional method and deep learning.Traditional methods are relatively easy to implement and highly interpretable,but the existing methods are difficult to deal with 2D plane information and inter-layer deformation simultaneously,and the interpolation effect is not ideal.Deep learning methods are usually based on optical flow or a cascade of image interpolation and deblurring.However,these methods rely on the quality of optical flow,the input image,and deblurring,making it difficult to produce accurate results when faced with challenges.Based on the above problems,this paper proposed the following two methods:(1)We proposed a medical image inter-layer interpolation network based on residual and inter-layer information fusion.Firstly,a cross residual generation network is proposed,in which multi-level skip connections are used to reduce the sensitivity of the model to the network`s depth,and a deeper network is constructed to make better use of the overall features of the image on the 2D plane and generate intermediate results.Then,an inter-layer information fusion module is constructed to extract compelling inter-layer features from the input images and integrate them into the intermediate results to supplement and enrich them.Finally,the weighted fusion loss function is used to measure the inter-layer image to accurately generate the inter-layer image.Experiments show that our method can obtain higher-quality inter-layer images on various tissue data sets.(2)We proposed a medical image slice interpolation tower network based on attention mechanism.Specifically,the tower interpolation module is used to expand the indirect domain of the image layer of the network to extract more image features.Secondly,the inter-layer structure information fusion module is used to process part of the input and the sub-output of the tower interpolation module to improve the motion consistency and complete the inter-layer interpolation.The above two modules enable the network make full use of the inter-layer deformation law and boundary contour features of the image.Furthermore,by introducing visual attention mechanism,we can accurately generate delicate tissue and texture elements.The experimental results show that the proposed method can accurately generate the pixels at the contour boundary and delicate tissue of inter-layer medical image.The experimental results on public data sets and self built data sets show that this method is superior to several advanced comparison methods.