阅读说明：本技术 一种基于图像识别判断结肠镜检查是否完整的检测方法 (Detection method for judging completeness of colonoscopy based on image recognition ) 是由 王玉峰 于 2019-04-25 设计创作，主要内容包括：本发明公开一种基于图像识别判断结肠镜检查是否完整的检测方法,包括：在整个肠镜检查过程中,利用完整度检测模型对肠镜视频流的每帧图像的四个角进行平均灰度值检测,若有一个角连续预定数量帧或以上输出的布尔值为0,则输出没有检查完整的提示。本发明的基于图像识别判断结肠镜检查是否完整的检测方法,可以解决医生在结肠镜检查过程中对肠道内部检查不完整的问题,从而提高操作者的结直肠镜检查质量,降低结肠镜检查病变漏诊率。(The invention discloses a detection method for judging whether colonoscopy is complete or not based on image recognition, which comprises the following steps: in the whole enteroscopy process, the four corners of each frame of image of the enteroscopy video stream are subjected to average gray value detection by using the integrity detection model, and if the Boolean value output by one corner for continuously presetting number of frames or more is 0, a prompt that the integrity is not checked is output. The detection method for judging whether the colonoscopy is complete based on image recognition can solve the problem that the doctor does not complete the examination of the intestinal tract in the colonoscopy process, thereby improving the quality of the colonoscopy of the operator and reducing the lesion missed diagnosis rate of the colonoscopy.)

1. A detection method for judging whether colonoscopy is complete or not based on image recognition is characterized by comprising the following steps: in the whole enteroscopy process, the four corners of each frame of image of the enteroscopy video stream are subjected to average gray value detection by using the integrity detection model, and if the Boolean value output by one corner for continuously presetting number of frames or more is 0, a prompt that the integrity is not checked is output.

2. The detection method for determining the completeness of colonoscopy based on image recognition according to claim 1, wherein in the whole colonoscopy process, the video stream transmitted from the colonoscope lens in the operating table is divided into two parts, one part is transmitted to the operating platform of the doctor, and the other part is sent to the completeness detection model for recognition after being subjected to frame preprocessing, so as to perform the detection of the completeness of the intestinal tract; and judging the brightness of four corners of the image according to the detection result by a threshold value, and outputting the detection result.

3. The detection method for determining whether the colonoscopy is complete based on image recognition as claimed in claim 2, wherein the physician is prompted on the physician operating platform that the colonoscopy is not complete when the detection result is incomplete.

Technical Field

The invention relates to the technical field of detection, in particular to a detection method for judging whether colonoscopy is complete or not based on image recognition.

Background

The incidence and mortality of rectal cancer in our country have shown a clear upward trend in recent years. Early discovery and early treatment are the main strategies for reducing the death rate of colorectal cancer at present. Enteroscopy is one of the most effective examination methods for finding colorectal polyps and tumors, and is also the main method for screening colorectal tumors. Colon polyps are found by using enteroscopy, and the detection result of the enteroscopy can be influenced due to the difference of intestinal tract preparation conditions of patients, the skill level of colonoscopy of operators, the shape and the size of the colon polyps and the like, and insufficient endoscope withdrawal time during detection. Adenomatous polyps serve as precancerous lesions, and the discovery rate of the adenomatous polyps is considered as one of the main indicators for measuring the quality of colonoscopy. Colonoscopy plays an important role in early detection, early diagnosis and early treatment of early cancer, and if missed detection occurs, the optimal time for treatment may be delayed, so that the death rate of patients can be reduced in a certain sense by reducing missed diagnosis to the maximum extent.

With the popularization of digestive endoscopy technology in China, the number of colonoscopy is increasing day by day. Deciding on colonoscopy is clearly a major psychological, economic and time challenge for the patient. Therefore, as a gastroenterologist and endoscopist, it is desirable to optimize the quality of a colon examination to maximize the benefit to the patient. It follows that the importance of the colonoscope quality control task goes undoubtedly.

However, in the existing colonoscopy process, due to different habits of each doctor and different operation techniques, the quality of the colonoscopy operation is uneven, and some doctors may not completely examine the whole intestinal tract due to lack of experience in the examination process.

The inability to perform a complete examination of the bowel during enteroscopy can pose a significant risk and result in missed detection of many diseases.

Disclosure of Invention

The invention aims to provide a detection method for judging whether colonoscopy is complete or not based on image recognition, aiming at the technical defects in the prior art.

The technical scheme adopted for realizing the purpose of the invention is as follows:

a detection method for determining whether a colonoscopy is complete based on image recognition, comprising: in the whole enteroscopy process, the four corners of each frame of image of the enteroscopy video stream are subjected to average gray value detection by using the integrity detection model, and if the Boolean value output by one corner for continuously presetting number of frames or more is 0, a prompt that the integrity is not checked is output.

In the whole enteroscopy process, a video stream transmitted by an enteroscopy lens in an operating table is divided into two parts, one part of the video stream is transmitted to an operating platform of a doctor, and the other part of the video stream is sent to an integrity detection model for identification after being subjected to frame preprocessing so as to carry out intestinal integrity detection; and judging the brightness of four corners of the image according to the detection result by a threshold value, and outputting the detection result.

In the detection process, in every continuous 200 frames, if the Boolean values of four corners are all 1, the number of checking turns is counted and added by 1.

Further, when the detection result is that the examination is not complete, the doctor is prompted to be not complete on the doctor operating platform.

Compared with the prior art, the invention has the beneficial effects that:

the detection method for judging whether the colonoscopy is complete based on image recognition can solve the problem that the doctor does not complete the examination of the intestinal tract in the colonoscopy process, thereby improving the quality of the colonoscopy of the operator and reducing the lesion missed diagnosis rate of the colonoscopy.

Drawings

FIG. 1 is a flow chart of the operation of the integrity detection model;

FIG. 2 is a flow chart of integrity testing throughout the enteroscopy testing procedure.

Detailed Description

The invention is described in further detail below with reference to the figures and specific examples. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

Fig. 1 shows a process of the integrity detection model, based on which the present invention determines whether the colonoscopy is intact using image recognition techniques.

The invention relates to a detection method for judging whether colonoscopy is complete or not based on image recognition, wherein an integrity detection model is operated by the following steps: step 1: detecting average gray values of partial areas of four corners in the enteroscope image;

step 2: data collection, wherein 800 samples in the four corner regions of 200 image screenshots in the enteroscopy process are divided into two types: brightness and darkness, wherein the two types of images are subjected to ambiguity detection through an MATLAB program respectively, test results are arranged in sequence, and a proper threshold value is found from the test results according to the proportion of a bright area to a dark area and is used for judging the brightness and darkness conditions of four corner areas;

and step 3: if the average gray value of one corner is smaller than the threshold value, displaying dark, and outputting a Boolean value of 0; if the average gray value of one corner is larger than the threshold value, displaying bright, and outputting a Boolean value 1; if the Boolean value of one corner in the continuous 10 frames of images is continuously output to be 0, judging that the direction of the corner is not completely checked;

And 4, step 4: a direction with a boolean value of 0 means that it was not detected for a while.

As shown in FIG. 2, the transmission process of the video stream in the whole enteroscopy process is realized by the following steps;

step 1: when an operation is started, a video stream transmitted by an enteroscope lens in an operating table is divided into two parts, one part is transmitted to an operating platform of a doctor, and the other part is transmitted to a integrity detection model for identification;

step 2: the method comprises the steps of preprocessing a video stream before artificial intelligence detection, and dividing the video stream into 25 frames per second to perform image integrity detection processing.

And step 3: judging the brightness of four corners of the image according to the detection result of the step 2 by a threshold value;

and 4, step 4: if the Boolean value of one corner in the continuous 10 frames of images is continuously output to be 0, judging that the direction of the corner is not completely checked;

and 5: in the detection process, in every continuous 200 frames, if the Boolean values of four corners are all 1, the number of checking turns is counted and added by 1.

Step 6: and if the direction is not detected, the doctor is prompted to be not detected completely on the doctor operating platform. And if the detection is complete, the subsequent artificial intelligence lesion detection is carried out, and the detection result is returned to the doctor operation interface.

The foregoing is only a preferred embodiment of the present invention, and it should be noted that, for those skilled in the art, various modifications and decorations can be made without departing from the principle of the present invention, and these modifications and decorations should also be regarded as the protection scope of the present invention.