Method for intelligently identifying and recording polyp extirpating video of endoscopic submucosal dissection

Abstract

The invention discloses a method for intelligently identifying and recording a polyp extirpating video of endoscopic submucosal dissection. The method comprises: capturing feature pictures of the operation video from the video in batches; establishing and training a neural network model; traversing the video again to verify the neural network model; producing prediction picture, comparing pictureswith relatively high similarity in the predicted pictures; archiving and storing the pictures without similarity in a data set; training the convolutional neural network again by using the archived data set; inputting the real-time endoscope video into the high-precision neural network model to be used for reading and classifying pictures in the operation process and recognizing specific operation instruments in the video so as to effectively record the video, and the method has the beneficial effect of being rapid and accurate in recognition.

Description

technical field [0001] The invention relates to the technical field of image recognition, in particular to a method for intelligently recognizing and recording a video of polyp removal by endoscopic submucosal dissection. Background technique [0002] Digestive endoscopy performs endoscopic submucosal dissection of the diseased part of the patient in the colon, that is, ESD surgery is a new treatment method that has emerged in recent years, and it is a technology with good clinical application prospects. The pain of abdominal surgery is characterized by small trauma, multiple treatments for the lesion, and low recurrence rate. However, the learning cost of the ESD surgery is relatively high, requiring doctors to observe a large number of surgical procedures and participate in practice. Easy access to operation videos, but video acquisition generally requires the cooperation of hospital and related department personnel. Today, domestic medical resources are unevenly developed...

AI Technical Summary

Problems solved by technology

[0002] Digestive endoscope performs endoscopic submucosal dissection on the patient's lesions in the colon, that is, ESD surgery is a new treatment method that has emerged in recent years. It is a technology with good clinical application prospects. The pain of abdominal surgery has the characteristics of small trauma, multiple treatments for lesions, and low recurrence rate. However, the learning cost of ESD surgery is high, and doctors need to observe a large number of surgical procedures and participate in practice. Recording surgical videos can make doctors more It is convenient to get access to the operation video, but the video acquisition generally requires the cooperation of the hospital and relevant department personnel. Today, the domestic medical resources are unevenly developed, and doctors and departments with the above-mentioned ESD operation ability have no time to perform operation video during their daily work. Recording is not conducive to the spread of ESD technology. In addition, the entire video recorded by the endoscope has a large storage capacity. If the video resolution is 1920*1080, the compressed video file size is about 20MB per minute, and a 30-minute surgical video needs to occupy 600MB. Storage space is also not conducive to proliferation. At the same time, the operation time is too long, and the actual operation time only accounts for a part of it. In fact, it is necessary to intercept the video of the operation part of the video to watch.

Images