Research Of Gastrointestinal Video Capsule Endoscopy Based On Wireless Power Transmission And Tumor Image Recognition

Wireless capsule endoscopy(WCE)is a convenient,noninvasive,and painless tool for detecting the whole digestive tract.It can get the most intuitive information of the lesions of human gastrointestinal(GI)tact.In recent years,researches on technologies concerning WCE have become popular in the field of medical devices at home and abroad.However,due to low image resolution,low frame rate and limited working time,the existing WCE systems cannot fully meet the requirements of clinical use.Meanwhile,each WCE examination produces tens of thousands images of GI tract.It is a tedious and time consuming task for a physician to review the WCE images one by one,which also hinder the application of WCE.Supported by many scientific research projects at the national and provincial level,this paper makes deeply research on the technologies of video capsule endoscopy based on wireless power transmission and the intelligent tumor detection of WCE images.Based on the the physiological features of the GI tract and its influence on the design of WCE system,two new kinds of video capsule endoscopy system prototypes have been designed and developed,including NTSC-based dual-head video capsule endoscopy system and Raw-based digital video capsule endoscopy system.Each functional module of the capsule has been studied and designed.New type of CMOS image sensor has been selected,and the matched pinhole lens and illumination module have been designed.According to the performance requirements,the micro control unit(MCU)is used to realize the design of control circuit and data processing module.According to the output formats of image acquisition module,the wireless video transmission circuit is proposed which has the advantages of high energy utilization,strong signal penetrability and wide bandwidth.Limited by the inner space of capsule,the helical antenna working in the normal mode is designed and the structure parameters are optimized by HFSS simulation software.The external video receiving system is proposed which could realize the video receiving,display,and storage.Meanwhile the real-time display function allows the doctors to monitor the status of video capsule endoscopy.In order to meet the energy requirements of video capsule endoscopy,the basic principle of inductive coupling energy transmission technology is studied to modularize the wireless energy transmission system under the security and stability constraints.The power transmitting circuit and elliptical solenoid pair transmitting coil are designed to satisfy the requirements of the safety of electromagnetic,the magnetic field uniformity and the frequency stability.The electromagnetic parameters like body tissue current density and SAR are obtained by building high accuracy body electromagnetic model.The magnetic field distribution functions of different transmitting coils are obtained from its original definition.The drop of driving current is evaluated to analyze the frequency drift of transmitting coils.The power receiving circuit and 3d hollow cylinder receiving coil are designed to satisfy the temperature rise safty and postural stability of receiving power.The safty range of internal resistance of receiving coil related to a load is obtained and the postural stability of receiving coil is studied based on Cartesian coordinate system.The power transfer efficiency is analyzed to make convenience to optimization design of the transmitting and receiving coils.In order to relieve the burden of physicians and improve the diagnostic rate,a novel automatic tumor recognition algorithm in WCE images has been proposed based on multi-scale analysis and fractal technology.In RGB and HSV color spaces,WCE textural images are synthesized by the inverse curvelet transform of the curvelet transform contained the appropriate scales for different color channels.Local fractal dimensions of the synthesized images are computed,and then,Gaussian second order moments and non-gaussian third order moments of textural measures extracted from the local fractal dimension matrices are computed to discriminate tumor and normal WCE images.The feature selection method and classifier based on genetic algorithm(GA)and support vector machine(SVM)are built to reduce the dimension of feature vector and select the optimal feature set owned the best classification performance.The parameters of SVM are optimized using K-fold cross-validation method to achieve the recognition of WCE tumor images.For the selected WCE images data,the automatic tumor recognition algorithm has been tested.The average classification rates reach up to 97.8% of sensitivity and 96.7% of specificity,which achieves the preliminary diagnosis of the small bowel tumor.Based on experiments study,the wireless powered video capsule endoscopy systems have been integrated and verified.The working performance of light source module,video capturing module,image processing module,and wireless communication module have been tested.In the in vitro experiment,the images of intestinal tract are captured clearly,which verified the entire function modules of the capsule endoscopy.In the wireless power transmission experiment,the power transmitting and receiving part have been respectively optimized through the experiments study.The postural stability of the new hollow cylinder receiving coil is proved higher than that of 3D orthogonal coil.Then,the experiments of wireless powered NTSC video capsule endoscopy system are conducted in different medium,which further prove the feasibility and the reliability of the power transmission and wirelss video communication under the influence of biological tissues.The phased objectives of the research project have been achieved,which lay the foundation for the further development of video capsule endoscopy.