Design And Implementation Of Digital Pathology Annotation Data Collection And Management System Based On Web

The development and maturity of image acquisition and storage technology not only reduces the cost of image acquisition and application,but also promotes the application and development of image processing and recognition technology in various fields.The application of image analysis and recognition technology relies on the construction of a specific image recognition algorithm,and the implementation and optimization process of the image recognition algorithm requires a large number of images with annotation information as a data set.The quality and quantity of the data set can directly affect the implementation of the image algorithm.In the digital pathology image annotation data collection process,the characteristics of large image size,high resolution,and low number of people with professional knowledge lead to low efficiency of data collection.Therefore,an effective method for solving the difficulties existing in the existing annotation data collection process is to improve the collection efficiency of the annotation data by designing and optimizing the annotation data acquisition system.In the existing image annotation data acquisition system,some are general annotation tools,and such annotation tools do not support the display of large-scale,high-resolution biomedical images.The others are specially designed tools for biomedical image display and annotation.Most of these tools are stand-alone software.T Its operation depends on a variety of supporting library files and environments.Besides,most of the stand-alone software only supports the display and storage of local images and annotation data.In view of the shortcomings in the existing labeled data acquisition system,this study uses Web technology to design and implement a software system based on browser/server structure for the annotation data collection and management of large-scale,high-resolution digital pathology images.The software consists of five parts: client,application server,data storage part and graphics processor unit server.According to the running device media and its functional characteristics,the system client can be divided into two versions: personal computer and mobile.Through the personal computer client of the system,the user can realize the basic functions of browsing digital pathological images and collecting data of regional annotations.With the shortcut module in the system,users can use shortcut keys instead of mouse to improve annotation efficiency.Besides,the user can complete the collection of the classification annotation of image patches through the mobile client of our system,and use the testing,history record querying and picture collection function module for review and discussion.On the basis of realizing the above basic functions,various management strategies are introduced in the system to independently manage images,annotation data and user information,and graphics processor servers are used to participate in the generation and management of annotation data.The application of the basic functions and management strategies in the above system effectively improves the efficiency of the annotation data acquisition on large-scale biomedical images,and can help managers maintain the quality of annotation data.Combined with the design of image-assisted interpretation software for clinical cervical liquid-based cytology,the PC client of this software system has been applied to collecting regional annotations data for more than 1 year,and the mobile terminal has also been used for classification annotation data collection for more than half a year.Good results have been achieved in improving the efficiency of the annotation data collection process.