CN-121999981-A - Rectus varus suture auxiliary system for fusion image recognition

CN121999981ACN 121999981 ACN121999981 ACN 121999981ACN-121999981-A

Abstract

The invention relates to the technical field of image analysis and identification, in particular to a rectus varus suture auxiliary system for fusion image identification, which comprises a permeation quantification and analysis module, a tissue state association analysis module, a risk feature learning module, a real-time feature analysis module, a risk evaluation and correction module and a comprehensive correction risk value and a compensation quantity recommendation value, wherein the permeation quantification and analysis module is used for calculating key bubble change performance parameters based on leaked bubble feature parameters to determine the activity degree of leakage concerned, the tissue state association analysis module is used for determining local tissue state parameters and local tissue significant parameters related to leakage, the risk feature learning module is used for generating local tissue significant feature vectors based on the local tissue significant parameters to generate a needle risk feature learning vector library, the real-time feature analysis module is used for extracting current local tissue significant parameters and local varus form quantification parameters, and the risk evaluation and correction module is used for evaluating basic risk values of current suture needles. The method improves the accuracy of risk early warning of the penetration of the anastomotic stoma in the auxiliary process of the rectal inversion suture.

Inventors

XING JUNJIE
YU GUANYU
WANG CHENGLONG
ZHAO ZIYE
Kang Zhengchun
WANG ZHEN
HAN XIANGLING
CHEN BINGCHEN

Assignees

中国人民解放军海军军医大学

Dates

Publication Date: 20260508
Application Date: 20260126

Claims (10)

1. A fused image-identified rectal inversion suture assist system, comprising: the seepage quantification and analysis module is used for calculating key bubble change performance parameters based on bubble characteristic parameters extracted from a plurality of image sequences of leakage occurring in a gas injection test in a historical rectal inversion suture operation video so as to determine leakage activity, and determining concern leakage activity based on the leakage activity; wherein the bubble characteristic parameters comprise bubble space density, track consistency and equivalent diameter; The tissue state association analysis module is used for determining local tissue state parameters corresponding to each suture needle and local tissue significant parameters related to leakage of each suture needle; the local tissue state parameters are extracted based on image sequences of the completion frames of the suture needles in the suture process corresponding to the leakage activity degree of interest, and the local tissue significant parameters are determined according to correlation analysis of the leakage activity degree of interest and the local tissue state parameters; the risk feature learning module is used for generating local tissue salient feature vectors of all suture needles based on the local tissue salient parameters and performing cluster analysis based on the acquired several local tissue salient feature vectors so as to generate a needle point risk feature learning vector library; The real-time characteristic acquisition module is used for acquiring corresponding suture needle completion frame images in the current rectal inversion suture operation; The real-time characteristic analysis module is used for extracting a current local tissue significant parameter and a local varus shape quantization parameter based on a frame image of each suture needle of the current rectal varus suture operation and generating a current local tissue characteristic vector of each suture needle based on the current local tissue significant parameter, wherein the local varus shape quantization parameter comprises local varus depth and transition smoothness with the last needle point; And the risk assessment and correction module is used for assessing the basic risk value of the current suture needle based on the similarity of the current local tissue feature vector and each vector in the needle point risk feature learning vector library, correcting the basic risk value based on the local inversion form quantization parameter, and generating the comprehensive correction risk value of the current suture needle and the recommended value of the next needle local lifting compensation quantity.
2. The fused image identified transrectal suture aid system of claim 1, wherein the permeation quantification and analysis module calculates key bubble variation performance parameters based on the bubble feature parameters; Wherein the key bubble variation performance parameters comprise bubble dynamic generation rate and bubble continuous growth ratio; The bubble dynamic generation rate is calculated according to a time sequence of bubble space density, the bubble continuous growth ratio is calculated according to an equivalent diameter sequence, and the equivalent diameter sequence is determined according to track continuity.
3. The fused image identified transrectal suture aid system of claim 2, wherein the permeation quantification and analysis module performs a weighted summation to determine leakage liveness based on the key bubble variation performance parameters.
4. The fused image identified transrectal suture aid system of claim 3, wherein the penetration quantification and analysis module identifies leakage of interest based on a determination that the leakage activity is greater than a leakage of interest activity threshold; wherein the leakage activity threshold of interest is determined based on a descending sequence of the leakage activity and a preset sequence percentile threshold.
5. The fused image identified transrectal suture aid system of claim 4, wherein the tissue state association analysis module extracts the local tissue state parameters based on an image sequence of each suture needle completion frame of the leakage liveness of interest corresponding suture process; wherein the local tissue state parameters comprise color expression parameters and texture expression parameters of a mucous membrane layer area and a serous layer area; the color expression parameters comprise a tone average value, a variance and a saturation average value; the texture performance parameters include gray matrix energy and local binary pattern uniformity.
6. The fused image identified transrectal suture assist system of claim 5, wherein the tissue state association analysis module compares the first correlation coefficient and the second correlation coefficient corresponding to each local tissue state parameter with a preset correlation coefficient threshold to determine a local tissue saliency parameter; The first correlation coefficient is determined by calculating the standard deviation corresponding to each local tissue state parameter of all suture needles respectively so as to calculate the correlation coefficient of the standard deviation and the leakage activity of interest; And the second correlation number is determined by respectively calculating the proportion of the suture needles of each local tissue state parameter which is larger than a corresponding preset interval so as to calculate the correlation coefficient of the proportion of the suture needles and the key leakage activity.
7. The fused image identified rectal inversion suture assist system of claim 6, wherein the risk feature learning module performs a cluster analysis based on the obtained plurality of local tissue saliency feature vectors to determine a plurality of cluster centers, and determines a plurality of needle risk feature learning vectors based on the local tissue saliency feature vectors corresponding to the plurality of cluster centers to construct a needle risk feature learning vector library.
8. The fused image identified rectocele suturing assistance system of claim 7, wherein the risk assessment and correction module is configured to calculate a similarity of a current local tissue feature vector of each suture needle to each vector in the needle point risk feature learning vector library to determine a base risk value based on a maximum of the similarities.
9. The fused image identified transrectal suture aid system of claim 8, wherein the risk assessment and correction module determines a comprehensive corrected risk value for a current suture needle based on a morphological correction coefficient and the base risk value; And calculating a depth deviation factor and a smoothness risk factor by the morphological correction coefficient based on the local varus morphological quantization parameter.
10. The fused image identified rectal inversion suture assist system of claim 9, wherein the risk assessment and correction module determines a next needle local pull-up compensation amount recommendation value based on the deviation amount of the local inversion morphology quantization parameter under a first preset condition; The first preset condition is to determine triggering compensation quantity recommendation based on comparison between the comprehensive correction risk value of the current suture needle and a preset comprehensive correction risk value threshold.

Description

Rectus varus suture auxiliary system for fusion image recognition Technical Field The invention relates to the technical field of image analysis and identification, in particular to a rectal inversion suture auxiliary system for fusion image identification. Background With the popularization of minimally invasive surgery and the development of computer-aided techniques, the application of intelligent methods such as image recognition and machine learning to surgical navigation and quality assessment has become an important trend for improving the accuracy and safety of surgery. In the field of colorectal surgery, in particular in low-grade anterior resection of rectal cancer, the quality of the rectal inversion suture is directly related to the occurrence of serious complications of postoperative anastomotic leakage as a key step in the reconstruction of the anastomotic stoma. Currently, studies have explored the use of endoscopic images for needle counting, gauge margin measurement, or visual guidance by overlaying pre-planned paths with augmented reality techniques that improve visualization and normalization of surgery to some extent. However, the prior art approaches focus on suture geometric parameters such as gauge, margin assessment or static path guidance based on preoperative planning, and fail to deeply consider two fundamental intrinsic relationships that affect anastomotic stoma healing, namely, the relationship between the intraoperative real-time tissue microscopic state and the postoperative leakage risk, and the physical relationship between the dynamically formed three-dimensional inversion morphology and the mechanical sealability of the anastomotic stoma during suturing. This results in a hysteresis in the risk assessment of anastomotic leakage and limited accuracy of assistance. CN119107474A discloses an anorectal operation suture effect evaluation method and system based on image recognition, and the anorectal operation suture effect evaluation method comprises the steps of obtaining treatment scheme data of anorectal operation, extracting suture feature influence parameter sets from the treatment scheme data, inputting the suture feature influence parameter sets into a pre-trained standard suture feature contrast model to generate a standard suture feature set, obtaining an operation suture image in real time, dividing and extracting suture position sub-images from the operation suture image, and inputting suture position sub-images into a pre-constructed suture feature image recognition model. It follows that the following problems exist in the prior art: The inherent relation between the microscopic state of the tissue and the leakage risk after operation and the mechanical sealing performance of the stitching three-dimensional form and the anastomotic stoma are not considered, the real-time risk assessment of each needle in the stitching process is not considered, and the quantitative operation guidance based on the risk and the form deviation is not considered, so that the problems of risk early warning hysteresis and low auxiliary assessment precision of the anastomotic stoma leakage are caused. Disclosure of Invention Therefore, the invention provides a rectal inversion suture auxiliary system for fusion image recognition, which is used for solving the problems that in the prior art, the internal connection of a tissue microscopic state and postoperative leakage risk and the internal connection of a suture three-dimensional form and mechanical sealability of a anastomotic stoma are not considered, real-time risk assessment is not considered for each needle in the suture process, quantitative operation guidance is provided based on risk and form deviation, and the risk early warning lag and auxiliary assessment precision of the anastomotic stoma leakage are low. To achieve the above object, the present invention provides a rectal inversion suture assist system for fusion image recognition, comprising: the seepage quantification and analysis module is used for calculating key bubble change performance parameters based on bubble characteristic parameters extracted from a plurality of image sequences of leakage occurring in a gas injection test in a historical rectal inversion suture operation video so as to determine leakage activity, and determining concern leakage activity based on the leakage activity; wherein the bubble characteristic parameters comprise bubble space density, track consistency and equivalent diameter; The tissue state association analysis module is used for determining local tissue state parameters corresponding to each suture needle and local tissue significant parameters related to leakage of each suture needle; the local tissue state parameters are extracted based on image sequences of the completion frames of the suture needles in the suture process corresponding to the leakage activity degree of interest, and the local tissue significant parameters are dete