Study On Automatic Morphological Examination Of Bladder Exfoliated Cells In Urine

The incidence of bladder cancer is a considerable one,and it has seen a steady rise in recent years.Presently,it is hard to diagnose it early.Urine exfoliative cytology screening can quickly and easily screen out early bladder cancer cells,improve the diagnostic sensitivity of early bladder cancer,and contribute to early diagnosis,treatment and follow-up monitoring.However,the traditional manual microscopy has high professional requirements,high labor intensity,and limited subjective factors,limiting the clinical use of this diagnostic method.Starting from the practical application of clinical auxiliary diagnosis,this paper studies the key technologies of high-quality image acquisition,exfoliated cell segmentation,recognition and other key technologies for the morphological examination of urinary bladder exfoliated cells,and developed a fully automatic microscopic examination instrument suitable for clinical application.Based on the micro-imaging system,the modular structure of the whole machine was designed,and the micro-vision system for automatic inspection of morphology of bladder exfoliated cells in urine was completed.The optical imaging of the system has good real color reducibility,and the mechanical structure has good and stable transmission accuracy,which lays a hardware foundation for collecting and recognizing cell images.In the aspect of automatic focusing,aiming at the phenomenon that there are many impurities in the medical pathology slides of urine based exfoliated cells and the evaluation of front and back images has multiple peaks,a comprehensive evaluation method for the sharpness of fusion image content pre-processing evaluation was proposed,which solved the problem of multi-peak discrimination in image sequence evaluation,and improved the image quality and efficiency of single-point focusing;During the whole scanning,the single-field focusing method of the slide based on the traditional mountain-climbing method and the full-field efficient automatic scanning strategy based on multi-point and fixed-plane were designed to greatly improve the scanning speed.The experimental results show that the fusion image content pre-processing comprehensive evaluation method combined with the motion acquisition synchronous background operation strategy has a focus success rate of 98.68%,a single point focus time of 936 ms,and a full slide scan time of 142 s for the 200 field of view,which meets the requirements of clinical liquid-based cell morphology and pathology automatic detection.In terms of segmentation of urinary-based bladder exfoliated cells,it combines color space theory,image preprocessing and image enhancement to achieve the requirements of background stripping and accurate segmentation of nuclei and plasmas of urinary-based exfoliated cells in complex environment.The H component was selected as the main spatial description of the model,and the weighted average gray component of the RGB color space was used as the supplementary description.The Otsu adaptive threshold segmentation method was used to extract the two color components and then take the union set to obtain the final exfoliated cell region;H component combined with dynamic threshold processing realized the extraction of nucleus and cytoplasm.In the aspect of classification and recognition,the multi-dimensional features of shape,color and texture were extracted by image segmentation algorithm to establish a morphological feature library,and support vector machines and BP neural networks with different kernel functions were constructed.Finally,the support vector machines with Gaussian Radial Basis Functions as kernel(RBF-SVM)were selected to realize the accurate identification of exfoliated cells,and a human-computer interaction system was built to complete the high-quality scanning acquisition,fast focusing,and The segmentation preview and recognition analysis function can assist manual work to promote the development of early screening and prevention technologies for bladder cancer and promote the advancement of intelligent medicine.The research results of this paper have been applied to the self-developed urine based exfoliated cell morphology microscanner.After the whole machine test system runs smoothly,the image acquisition is efficient and clear,and has the integrated functions of collection,segmentation,labeling,preview,recognition analysis and output of diagnostic reports.The comprehensive recognition rate of urine based exfoliated cells is 96.83%.Clinical can replace artificial microscopy,and promote the development of early screening and prevention technology for bladder cancer.