The Design And Implementation Of Digital Pathology Diagnostic Software System

With the development of computer technology,more and more medical staff realized the inefficiency of traditional microscopes and began to turn their attention to the direction of digital pathology,which greatly promoted the development of digital pathology and its remote diagnosis,and has contributed to a considerable industry outlook.To this end,through in-depth research on digital pathology related technology,this system has completed the core algorithms such as image focus,image registration algorithm,image analysis and other algorithm in the process of digitization of pathological slices.Designed and implemented a digital pathological intelligent diagnosis software system.The system can realize rapid scanning of pathological sections,seamless image splicing,and DNA intelligent analysis.The DNA analysis algorithm can achieve the purpose of auxiliary diagnosis and has strong practical value.The main innovations are as follows:1.In this paper,a CUDA-accelerated template registration algorithm was proposed.During the process of digitalization of pathological slices,the thousands of pathological images acquired were quickly registered and the problem of slow image registration speed was overcome.This algorithm uses GPU multi-threaded parallel computation to do hardware acceleration for template matching,which greatly improves the image registration efficiency and meets the requirements of real-time registration.2.In this paper,a weighted image fusion algorithm based on brightness adjustment is proposed to perform weighted fusion in the overlap region of the image,this algorithm can eliminates seams in the digitization process image mosaic which caused by inconsistent brightness of the captured image due to unstable exposure or other reasons.The weighted fusion algorithm can use the weights of different pixels to achieve a smooth transition of t he splicing region,and the brightness adjustment enables the spliced image to have a uniform brightness so as to achieve a good visual experience.3.An image stitching method based on weighted path planning is proposed,which overcomes the problem of image registration failure caused by the accumulation of registration errors during the stitching process of thousands of images.The core idea of weighted path planning is to prioritize the images with high registration rate,and generate the image mosaic path according to the registration size to prevent the error accumulation.4.In this system,based on the completion of pathological digitization,an intelligent analysis of pathological sections was also achieved.This paper proposes an image adaptive OTSU segmentation algorithm based on OpenMP acceleration to complete the segmentation of pathological cells.The image of the cell was identified by OpenCV SVM.After statistical calculations,relevant parameters of the suspected diseased cells are given and pathologists are provided to assist in the diagnosis.Finally,the diagnosis results are given by the doctor and the efficiency of the doctor is improved.