| The global female cancer mortality rate is second of the cervix cancer,which can be cured if it can be diagnosed early.The current clinical screening for cervical cancer are: by living tissue sampling in the white light image acquisition under the guidance of colposcopy,and confirmed the samples under light microscope after staining the pathological examination,so the whole process is not only cumbersome and time-consuming,but also excessive dependence on the doctor’s subjective experience.Therefore,optical pathology diagnosis technology combining spectral analysis and optical imaging has become one of the research hotspots nowadays.This technology is expected to become an objective,comprehensive,non-invasive and timely screening method for cervical cancer.However,the instruments formed by optical pathology diagnosis generally have the problems of large volume,unstable structure,low time efficiency,and few intelligent diagnostic functions.Therefore,this paper presents a rapid narrow band multispectral imaging method for screening cervical cancer.Based on the four characteristic bands of 415 nm,450nm,525 nm and 620 nm,the method is used to image the cervical tissue,and the acquired multispectral images are analyzed and processed.Based on that,a semi intelligent multispectral pathological diagnosis algorithm for cervical cancer is established.The main contents are as follows:(1)A micro fast narrow band multispectral imaging method is established.It includes: first,we integrate microfiltration and image sensor to form a miniature fast narrow band multispectral imaging module,build the whole imaging system,and achieve zero time difference to get four characteristic spectral images of cervical tissue quickly.Secondly,grid based ghost correction algorithm is studied.The correction algorithm can eliminate the mesh artifacts of the multispectral imaging module integrated with the fixed size microfiltration and any type of image sensor.Thirdly,the algorithm of fusion and reduction of near natural color images is carried out,which makes the fusion images of micro fast narrow band multispectral imaging close to natural color and maximizing the clinical observation habits of doctors.Finally,the establishment of cervical lesions classification algorithm of semi intelligent,a regional doctors only need to manually enter different types of organizations,this algorithm can have the entire lesion on the cervical tissue judgment,if there is disease belongs to the level of the lesion,and will give out the boundary contour lesions.(2)A preliminary clinical verification of the micro fast narrow band multispectral imaging method was carried out.The use of micro fast narrow band multi spectral imaging method for imaging and treatment of cervical cancer in vitro,and the final results of the classification of lesions and pathological examination(gold standard)results of objective comparison,and then handed over a number of experienced gynecologic experts subjective evaluation.The experimental results show that the method can effectively classify cervical tissue into the same type of lesion as the gold standard diagnostic result without the aid of iodine and acetic acid reagent.Especially in a variety of levels(CIN1,CIN2,CIN3,canceration),it is difficult for doctors to distinguish the area from the naked eye,and the edge of the lesion is difficult for doctors to define.This technology can play a good role in prompting and suggesting.Through the study of the above content,the main innovation is to achieve a miniature fast and narrow band multi spectral imaging of the cervical tissue;and the establishment of a noninvasive results instantly available cervical cancer pathological diagnosis in the optical method,reduce the dependence on the subjective experience of doctors,the intuitive,fast and accurate divide on the different grades of cervical tissue lesions.The research of this subject lays the foundation for the application of optical pathological diagnosis technology in clinical cervical cancer screening,and is of great significance for the popularization of cervical cancer screening and the precise treatment of cervical cancer. |