Research And Implementation Of Medical Image Fast Labeling System For Deep Learning Application

Deep learning is an important method to realize the development of artificial intelligence in medical image.At present,due to the lack of high-quality labeling data,the development of artificial intelligence in medical image is slow.Because of the specialty and complexity of medical image annotation,the simple manual annotation efficiency is difficult to meet the annotation data volume of deep learning,and the automatic image annotation is difficult to meet the accuracy of annotation requirements.Therefore,this paper uses the combination of image annotation and text annotation to solve the research process of medical image annotation with the semi-automatic annotation method of automatic text annotation + manual image annotation Research and develop an efficient,fast and semi-automatic annotation system.In view of the above key problems,this paper intends to build a medical image knowledge map,use the method of matching the image report text with the knowledge map after segmentation by natural language processing technology,and automatically generate the text annotation of structured text.By visualizing the text annotation,the annotation experts can clearly and quickly understand the image information and lesions to be labeled,and then Manual image annotation is performed on the desensitized medical image.After the system analysis,design,implementation and testing stage,the developed medical image rapid annotation system for deep learning can annotate the medical image data quickly and semi automatically.After experimental comparison and statistical analysis,the annotation method of medical image rapid annotation system is better than the traditional manual annotation method in the time of annotation.To a certain extent,the realization of the medical image rapid annotation system for deep learning can improve the annotation efficiency of the annotation experts for the image data,provide efficient and high-quality annotation data for the training and testing of deep learning in the medical image,and then promote the research of artificial intelligence in the field of medical image.