Search

CN-121074120-B - Artificial intelligence identification method for urodynamic examination image

CN121074120BCN 121074120 BCN121074120 BCN 121074120BCN-121074120-B

Abstract

The invention relates to the field of image analysis, in particular to an artificial intelligent identification method of a urodynamic examination image. The method comprises the steps of obtaining urodynamic contrast images and displacement information of monitoring points, calculating a speed time domain curve based on the displacement information, grabbing a steady-state time period, aligning and positioning key contrast frames through a time sequence, selecting adjacent continuous frames through a sliding window to construct a reference frame group, identifying tissue contours of the reference frame group to generate a superimposed thermodynamic diagram, determining a superimposed main contour through frequency threshold screening and contour closing processing, calculating interval width vertical to the contour based on a displacement extremum, constructing a strip-shaped dynamic reference interval along the main contour, and locally enhancing images in the tissue contour reference interval. The invention pointedly considers the interference of the actuation state of physiological motion, improves the development quality of a key area, improves the image processing efficiency, and is helpful for assisting doctors in improving the diagnosis speed and efficiency.

Inventors

  • XU BEN
  • LIU SHIHAO

Assignees

  • 北京大学第一医院(北京大学第一临床医学院)

Dates

Publication Date
20260508
Application Date
20250826

Claims (10)

  1. 1. An artificial intelligence recognition method of a urodynamic examination image, comprising the steps of: continuously acquiring urodynamic contrast images of the monitoring targets, and synchronously acquiring displacement information of the monitoring points corresponding to the monitoring targets; determining a monitoring point speed time domain change curve based on the displacement information so as to grasp a speed steady-state time point; aligning the time domain change curve of the speed with the starting time of the urodynamic contrast image, grabbing key contrast frames in the urodynamic contrast image based on a steady-state time point, and determining a plurality of adjacent contrast frames based on the key contrast frames to construct a reference frame group; Identifying the tissue contour of each contrast frame in the reference frame group, carrying out superposition processing on each contrast frame, determining a superposition main contour, determining a section width based on displacement information, and constructing a confidence tissue contour section based on the section width and the superposition main contour; The remaining contrast frames of the non-reference frame set in the urodynamic contrast image are enhanced based on the confidence tissue profile interval, including, Determining a confidence tissue contour interval in the rest contrast frames, and locally enhancing the image in the confidence tissue contour interval; the displacement information comprises coordinate data of the monitoring points at all moments.
  2. 2. The artificial intelligence recognition method of urodynamic examination image according to claim 1, wherein obtaining displacement information of the monitoring target corresponding to the monitoring point comprises, Acquiring coordinate data of a plurality of monitoring points of the monitoring target in a three-dimensional space in real time through a spatial position sensing device; the detection points comprise chest monitoring points and abdomen monitoring points.
  3. 3. The artificial intelligence method for identifying urodynamic examination images according to claim 1, wherein determining a monitoring point velocity time domain change curve based on displacement information to grasp a velocity steady state time point comprises, Calculating the instantaneous speed of each monitoring point at a plurality of moments based on the displacement information; Calculating the average value of the instantaneous speed of each monitoring point to generate a speed time domain change curve; And screening a steady-state time period in which the average value of the instantaneous speed is continuously lower than the speed threshold value based on the set speed threshold value, and determining the lowest speed point in the time period as the steady-state speed time point.
  4. 4. The artificial intelligence method for identifying a urodynamic examination image according to claim 1, wherein aligning the velocity time domain change curve with the urodynamic contrast image start time, capturing key contrast frames in the urodynamic contrast image based on the steady state time point comprises, Determining the time point of each frame image sequence of the urodynamic contrast image in the time domain change curve of the speed; and selecting a contrast frame which is closest to the speed steady-state time point in time as a key contrast frame.
  5. 5. The artificial intelligence method for identifying a urodynamic examination image of claim 3, wherein determining a plurality of contiguous contrast frames based on key contrast frames to construct a reference frame set includes, Based on the key contrast frames, a number of consecutive frames forward and backward of the key contrast frames are selected during the steady state period to construct a reference frame set.
  6. 6. The artificial intelligence method for identifying an image for urodynamic examination according to claim 1, wherein identifying a tissue contour of each contrast frame in the set of reference frames, the process of superimposing each contrast frame comprises, Image segmentation processing is carried out on each contrast frame in the reference frame group, and corresponding contours in each contrast frame are extracted; Aligning the contrast frame images after the contrast frame images are placed in the same space coordinate system; And calculating the frequency of covering each position point in space by the contour based on each aligned contrast frame image, and generating a contour superposition thermodynamic diagram.
  7. 7. The artificial intelligence method of identification of urodynamic examination images of claim 6, wherein the process of determining the superimposed main contour comprises, And setting a preset frequency threshold value, and screening out a contour with the coverage frequency exceeding the frequency threshold value from the contour superposition thermodynamic diagram as a superposition main contour.
  8. 8. The artificial intelligence method of identification of urodynamic examination images of claim 1 wherein constructing a confidence tissue profile interval based on the interval width and the superimposed primary profile comprises, Constructing virtual reference main contours at two sides respectively from the interval width by taking the superimposed main contours as references; and determining a banded region formed by the virtual reference main contour as a confidence tissue contour interval.
  9. 9. The artificial intelligence method for identifying urodynamic examination images according to claim 8, wherein determining the interval width based on the displacement information comprises, Determining an instantaneous speed average value of each detection point in a time domain segment corresponding to the reference frame group; And determining that the interval width is positively correlated with the instantaneous speed average value.
  10. 10. The artificial intelligence method of identification of urodynamic examination images of claim 1, further comprising identifying confidence tissue contour intervals in each remaining contrast frame.

Description

Artificial intelligence identification method for urodynamic examination image Technical Field The invention relates to the field of image analysis, in particular to an artificial intelligent identification method of a urodynamic examination image. Background Urodynamic examination requires comprehensive assessment of clinical symptoms and lower urinary tract functions, the main objective of which is to reproduce the symptoms of the patient and to determine the cause of the symptoms by urodynamic measurement or observation, and interpretation of urodynamic examination results should comprehensively consider the patient's medical history, including symptoms, combined diseases/conditions, and as much other information as possible, such as residual urine volume, urination diary, etc. For example, chinese patent publication No. CN101933812A discloses a urodynamic detection and analysis method, which comprises the steps of establishing an elastic element model of a bladder, establishing a urethra model, keeping the front urethra vertical to the gravity direction, recording and measuring urination data, calculating urination parameters, calculating the contraction length of the elastic element, further calculating the contraction speed, the contraction acceleration and the maximum contraction acceleration of the elastic element, and calculating the maximum cross-sectional area of the urethra model. The invention can completely overcome the pain and the infection possibility of the patient caused by the traditional invasive urodynamic examination method, the whole analysis process can be automatically carried out with the help of a computer, the result is clear and clear, the clinical memory and the use are convenient, and the device manufactured by the method has simple structure and convenient maintenance, thereby reducing the medical cost. However, the prior art has the following problems 1. In the prior art, dynamic interference caused by physiological displacement such as patient respiration and the like on urodynamic contrast images in the examination process is not considered. The method leads to systematic deviation in parameter calculation based on a static or quasi-static model, and can not accurately capture the real morphological change of the organ boundary in a dynamic scene, so that the subsequent doctor has higher evaluation and diagnosis difficulty and higher manual workload. 2. In the prior art, in the critical physiological phases such as the bladder filling phase, the urination phase and the like, the accuracy of identifying the tissue boundary in the image frame with serious motion artifact is deteriorated due to an inadaptive image processing mode, so that the subsequent doctor has higher evaluation and diagnosis difficulty and higher manual workload. Disclosure of Invention Therefore, the invention provides an artificial intelligent identification method of urodynamic examination images, which is used for solving the problems that in the prior art, dynamic interference on urodynamic contrast images caused by physiological displacement such as patient respiration and the like in the examination process is not considered, and an image processing mode of unadapted adjustment causes higher difficulty in subsequent evaluation and diagnosis of doctors and higher manual workload. In order to achieve the above object, the present invention provides an artificial intelligence recognition method of urodynamic examination image, comprising: continuously acquiring urodynamic contrast images of the monitoring targets, and synchronously acquiring displacement information of the monitoring points corresponding to the monitoring targets; determining a monitoring point speed time domain change curve based on the displacement information so as to grasp a speed steady-state time point; aligning the time domain change curve of the speed with the starting time of the urodynamic contrast image, grabbing key contrast frames in the urodynamic contrast image based on a steady-state time point, and determining a plurality of adjacent contrast frames based on the key contrast frames to construct a reference frame group; Identifying the tissue contour of each contrast frame in the reference frame group, carrying out superposition processing on each contrast frame, determining a superposition main contour, determining a section width based on displacement information, and constructing a confidence tissue contour section based on the section width and the superposition main contour; The remaining contrast frames of the non-reference frame set in the urodynamic contrast image are enhanced based on the confidence tissue profile interval, including, Determining a confidence tissue contour interval in the rest contrast frames, and locally enhancing the image in the confidence tissue contour interval; the displacement information comprises coordinate data of the monitoring points at all moments. Further, the displacement info