Research On Fusion Technology Of Multimodal Gastrointestinal Tumor Image Based On EUS

As we all know,image information has the characteristics of intuition,vividness and so on.This is so-called "Seeing is believing." In order to obtain the image information of the internal organs of the human body,ultrasound is a commonly used medical examination and diagnosis method undoubtedly,which is widely used in daily examinations in the gastroenterology,urology,obstetrics,gynecology and cardiovascular departments.It is estimated that gastrointestinal tumors,which are diseases of the digestive system,have become the most common malignant tumors in China.More than 1 million people die of esophageal cancer,gastric cancer and colorectal cancer every year.The most critical and simplest measure to improve the prognosis of gastrointestinal tumors is early treatment.As a result,the key measure to improve the prognosis of digestive tract tumors is early treatment.In view of this,this article relies on the National Key R&D Program "Digital Diagnostic and Medical Equipment R&D" key special project "Electromagnetic Acousto-optic Coupling Steep Pulse High Voltage Electric Field(s PEF)Non-invasive Gastric Cancer Treatment System Development"(Project Number: 2018YFC0115300),using the changes in the electromagnetic,optical and acoustic physical characteristics of gastrointestinal tumors as the entry point,and based on the ultrasound images of gastrointestinal tumors examined by EUS,the three-dimensional reconstruction of ultrasound images of gastrointestinal tumors and the fusion technology of three-dimensional ultrasound images and optical endoscopy images are carried out.The main contents are as follows:(1)On the basis of extensive research and documentation at home and abroad,it is determined that real-time evaluation of tumor ablation degree is a key factor to ensure the safety and effectiveness of tumor ablation.Based on this,in view of the lack of specific and clear indicators for biomorphological characterization during the ablation process,the study has analyzed and determined the morphological characteristics,location and boundary of gastrointestinal tumors,which is of very important and profound significance for the precise treatment of early cancer.(2)Given the unclear boundary due to the essential characteristics of ultrasound image,the false boundary due to physiological motion or operation orientation,the widespread high-level speckle noise and other negative effects,it is difficult to use existing methods to archive the 3D reconstruction of medical ultrasound images directly.Therefore,research work on image preprocessing such as ultrasound image filtering is carried out.An image filtering algorithm based on sparse representation is proposed,which can achieve good results in subjective evaluation and objective indicators like CNR for denoising of synthetic speckle images and actual medical ultrasound images.(3)In view of the problem that the feature information of the visible image and the ultrasound image does not match,the spatial information is asymmetric,and the direct fusion is difficult,fusion after 3D reconstruction is adopted.The characteristics of surface rendering and volume rendering are analyzed in detail,and ray casting is algorithm is selected to complete the reconstruction of ultrasound images.The results show that volume rendering reconstruction has more advantages.Compared with the unfiltered reconstruction effect,the three-dimensional image processed by the filtering algorithm proposed in this thesis can better display the image information.(4)For the 2D visible images obtained during common gastrointestinal endoscopy cannot provide more useful anatomical information,the limitations of a single modality and the multi-modality image are explained.Complementarity,through multi-modal image registration and fusion to explore the intrinsic correlation of multi-source biological information,the principle and technical method of registration fusion 2D visible image and 3D ultrasound image are constructed.That is,the two different images are registered by creating a 2D ultrasound projection from the reconstructed 3D ultrasound volume and registering it to the visible image.After the registration,the visible image is projected onto the 3D ultrasound volume to complete the fusion,and then use subjective evaluation to evaluate it.The results show that combing tumor biological information will make the fusion of visible images and ultrasound images more effectively,which can determine the extent of tumor invasion from the two dimensions,width and depth.It can not only achieve multi-angle and multi-level positioning and analysis of tumor lesions,but also provide reliable and accurate auxiliary judgment information for tumor staging and tumor ablation effect.