| Wireless capsule endoscope(WCE)is a painless,safe,and convenient detection tool for gastrointestinal diseases.By collecting the images of the inner wall of gastrointestinal tract,it provides the most intuitive information about lesions for clinicians.It is huge progress to noninvasive and early detection technology of gastrointestinal disease.However,its size is too tiny to contain sufficient batteries.As a result,it works less than 10 hours and does not cover a whole digestion period,causing missed diagnosis for some certain locations.Moreover,novel functions like high-resolution imaging,active locomotion of lingering,and biopsy are promoted in recent years,which rises its power consumption further.Power supply turns out to be a serious bottleneck restricting the development of WCE.Furthermore,hundreds of thousands of gastrointestinal images are captured in a complete WCE detection.It is too time-consuming and infeasible for clinicians to manually recognise and diagnose all images.Besides,the current principle for disease recognition is based on morphological and color of tissues,which is difficult to discover tiny anomaly of precancerous lesion or early cancer.In this paper,under the support of national and provincial scientific research foundations,key technologies of noninvasive capsule endoscope system(NCES)for gastrointestinal tumor and autofluresecence imaging system(AFIS)for precancerous lesions are carried out.Power supply is the prerequisite for normal work of NCES.The imaging capsule of NCES is in patient's intestine so that it cannot be powered by external electrical source with wires.And its size is too tiny to contain batteries with large capacity.So it should be supplied via wireless power transmission technology.The position and angle of power receiving coil in gastrointestine are always changing relative to extracorporeal transmitting coil,which leads to variation of receiving power for imaging capsule.If the receiving power is insufficient,the NCES cannot work normally.So it is necessary to analyze the mutual inductance between the transmitting and receiving coils when the latter is in all possible positions and angles,so as to obtain the distribution of receiving power.A coupled mutual inductance model for relative positional and angular misalignments is proposed: according to the coil winding parameters,from single turn to multiple turn and then to multiple layers;according to the positional relationship,from coaxial to parallel but not coaxial to generally non-parallel situation.Compared with existing analytical methods and finite element simulation methods,the advantages of this model in accuracy and consumption time are obtained.The measurement platform is built to measure the mutual inductance of each position and angle,and compared with calculation results of the model.The accuracy of the model is further verified.The mutual inductance when receiving coil is within biological tissues is measured to study the influence of electromagnetic absorption effect of biological tissues on the model.The control variable method is used to analyze the effect the axial,radial,and angular misalignment on mutual inductance.Receiving coils with different geometry parameters are wound to explore the influence of coil parameters on mutual inductance.The model is not only applicable to NCES,but also many other medical microdevices utilizing wireless power transmission technology,for which the admissible condition is proposed.Receiving coil with novel structure is proposed to solve the energy shortage,and also reduce the volume of imaging capsule.Based on constraints from actual environment of gastrointestinal tract and safety issue of heat dissipation on human tissues,optimal design parameters such as the number of turns and the shape of the receiving coil are obtained,so that the receiving power meet the least demanding power,and it should be sufficient and stable at as many angles and positions as possible.The optimal receiving coil is equipped into NCES to implement imaging experiment on live pig.Images are clear with low noise,and the video stream is smooth and stable.It turns out that NCES works normally and the coil provides sufficient and stable energy for imaging capsule.In a complete detection by NCES,more than 100,000 images are captured.It is infeasible for clinicans to manually diagnose all of them one by one,and visual fatigue happens after reading so many highly similar images,leading to misdiagnosis and missed diagnosis.In this paper,a method of classification and recognition for tumor image is proposed based on human-computer cooperation,which aims to reduce the false negative rate and identify all positive images.It works as following.Firstly,process is run in computer: color space is converted to YCb Cr;curvelet transform is performed to get sub images in multiple scales,angles,and positions in frequency domain;then select the best sub images according to the fitness function with differential of lacunarity in genetic algorithm;extract the global and local texture features;and input the feature vector into radial basis function neural network for image preliminary classification.Then,for "suspected positive" images,the final diagnosis is made after additionally manual check by clinicians.The clinician-computer cooperation reduces the false negative rate effectively,so as to prevent positive patients from being misdiagnosed as healthy and missing the golden treatment period.The performance of the proposed method is verified by standard data set firstly;accuracy and feasibility are further validated by images captured by NCES.At present,it is difficult to detect early cancers and precancerous lesions.The stage of cancer is decisive for the treatment plan and prognosis for patients.The medical treatment for intermediate and terminal cancers is difficult and prognosis is poor.Early diagnosis of precancerous lesions can save many patients' lives.Based on the fact that the density of intrinsic fluorescent substance in precancerous tissues is different from that of normal tissues,these different kinds of tissues produce different autofluorescence images after excitation of ultraviolet(UV)light source.AFIS is proposed to detect precancerous lesions.Aiming at the problems of overheating and beam divergence of UV LED,the heat dissipation and light condenser devices are designed respectively.The white light and AFIS images of three types of tissues(normal tissue,benign hyperplasia and precancerous lesions)are compared with each other.Obviously,AFIS distinguishes three kinds of tissues more clearly,and it does not get affected by shadow of fold and other morphological factors of tissues. |
PDF Full Text Request