CN-122024237-A - Intelligent aortic dissection rupture risk identification method and system based on semantic segmentation

CN122024237ACN 122024237 ACN122024237 ACN 122024237ACN-122024237-A

Abstract

The invention relates to the technical field of image data processing, in particular to an intelligent identification method and system for aortic dissection fracture risk based on semantic segmentation, comprising the steps of collecting CT images of the same patient at different time points; the method comprises the steps of obtaining a plurality of connected domains of an aortic region in a CT image, obtaining cavity wall morphology irregularity of the cavity wall connected domains, determining a noise cavity region based on the cavity wall morphology irregularity, obtaining a non-fracture prediction coefficient by using the true and false connected domains of the noise cavity region, determining a risk trend index according to the non-fracture prediction coefficient, determining the noise cavity region with fracture risk according to the risk trend index, obtaining a risk prediction coefficient of the noise cavity region with fracture risk based on the risk trend index of the noise cavity region with fracture risk and the cavity wall morphology irregularity, and obtaining a risk level assessment result according to the risk prediction coefficient. The risk identification method and the risk identification device can improve the accuracy of risk identification.

Inventors

WU TAO
LI FENGHUA
HUANG JIA

Assignees

延安大学咸阳医院有限公司

Dates

Publication Date: 20260512
Application Date: 20260203

Claims (10)

1. The intelligent identification method for the aortic dissection rupture risk based on semantic segmentation is characterized by comprising the following steps of: Acquiring CT images of the same patient at different time points, wherein one time point corresponds to one CT image; Carrying out semantic segmentation on the CT image to obtain a plurality of connected domains of an aortic region in the CT image, wherein the plurality of connected domains at least comprise a true cavity connected domain, a false cavity connected domain and a cavity wall connected domain; acquiring the cavity wall morphology irregularity of the cavity wall communicating domain by using the Hough circle detection result of the cavity wall communicating domain, the edge pixel points and the geometric center of the cavity wall communicating domain; determining a noisy cavity region based on cavity wall morphology irregularities; acquiring a non-fracture prediction coefficient of the noise cavity region by using the true and false connected region of the noise cavity region; determining a risk trend index of the noise cavity region according to the non-fracture prediction coefficient of the noise cavity region; Determining a noise cavity region with a cracking risk according to the risk trend index of the noise cavity region; acquiring a risk prediction coefficient of the noise cavity region with the cracking risk based on the risk trend index of the noise cavity region with the cracking risk and the cavity wall morphology irregularity; and acquiring a risk level evaluation result according to the risk prediction coefficient of the noise cavity region with the rupture risk.
2. The intelligent recognition method for aortic dissection fracture risk based on semantic segmentation according to claim 1, wherein the acquiring of the cavity wall morphology irregularity of the cavity wall connected domain by using the hough circle detection result of the cavity wall connected domain, the edge pixel point and the geometric center of the cavity wall connected domain comprises the following specific steps: Calculating the absolute value of the difference between the abscissa of the edge pixel point and the abscissa of the geometric center of the cavity wall communicating domain and the absolute value of the difference between the ordinate of the edge pixel point and the ordinate of the geometric center of the cavity wall communicating domain for each edge pixel point of the cavity wall communicating domain, and summing the two absolute values of the difference to obtain the difference between each edge pixel point of the cavity wall communicating domain and the geometric center; Summing the difference values of all edge pixel points of the cavity wall connected domain and the geometric center to obtain the difference value of the cavity wall connected domain and the geometric center; Calculating the ratio of the side length of the cavity wall communicating domain to the Hough circle detection result of the cavity wall communicating domain, and multiplying the ratio by the difference value between the cavity wall communicating domain and the geometric center to obtain the cavity wall morphology irregularity of the cavity wall communicating domain.
3. The intelligent identification method for aortic dissection fracture risk based on semantic segmentation according to claim 1, wherein the determining of the noisy cavity area based on cavity wall morphology irregularity comprises the following specific steps: and when the normalized cavity wall morphology irregularity of the cavity wall connected domain is larger than or equal to a preset irregularity threshold value, judging the cavity wall connected domain as a noisy cavity region.
4. The intelligent identification method for aortic dissection fracture risk based on semantic segmentation according to claim 1, wherein the obtaining of the non-fracture prediction coefficient of the noisy cavity region by using the true and false connected domain of the noisy cavity region comprises the following specific steps: and for the real cavity connected domain and the false cavity connected domain corresponding to each noise cavity region, calculating the ratio of the real cavity connected domain area to the false cavity connected domain area, and the average value of the Hough circle detection result of the real cavity connected domain and the Hough circle detection result of the false cavity connected domain, and determining the product of the ratio and the average value as the non-fracture prediction coefficient of each noise cavity region.
5. The intelligent identification method for aortic dissection rupture risk based on semantic segmentation according to claim 1, wherein the determining the risk trend index of the noisy cavity region according to the non-rupture prediction coefficient of the noisy cavity region comprises the following specific steps: acquiring non-fracture prediction coefficients of each noise cavity region at different time points, and forming a non-fracture prediction coefficient sequence of each noise cavity region according to the time point sequence; And calculating the ratio of two adjacent elements for the non-fracture prediction coefficient sequence of each noise cavity region, summing the ratio of all the two adjacent elements, and dividing the sum by a first quantity to obtain a risk trend index of each noise cavity region, wherein the first quantity is the element quantity minus 1.
6. The intelligent identification method for aortic dissection rupture risk based on semantic segmentation according to claim 1, wherein the determining the noise cavity region with rupture risk according to the risk trend index of the noise cavity region comprises the following specific steps: And when the risk trend index of each noise cavity area is larger than a preset index threshold value, judging that the noise cavity area has a cracking risk.
7. The semantic segmentation-based aortic dissection fracture risk intelligent identification method according to claim 6, further comprising: And when the risk trend index of each noise cavity area is smaller than or equal to a preset index threshold value, judging that the noise cavity area is free from cracking risk.
8. The intelligent identification method for aortic dissection rupture risk based on semantic segmentation according to claim 1, wherein the acquiring the risk prediction coefficient of the noisy cavity region with rupture risk based on the risk trend index and the cavity wall morphology irregularity of the noisy cavity region with rupture risk comprises the following specific steps: And for the noisy cavity area with the rupture risk, calculating the average value of the morphological irregularity of the cavity wall of the noisy cavity area at different time points, and determining the product of the average value and the risk trend index of the noisy cavity area as the risk prediction coefficient of the noisy cavity area.
9. The intelligent identification method for aortic dissection rupture risk based on semantic segmentation according to claim 1, wherein the step of obtaining a risk level assessment result according to a risk prediction coefficient of a noise cavity region with rupture risk comprises the following specific steps: When the risk prediction coefficient of the noise cavity region with the rupture risk after normalization is in a preset first threshold value interval, judging that the risk level is lower, and needing to be monitored periodically; When the risk prediction coefficient of the noise cavity region with the rupture risk after normalization is in a preset second threshold value interval, judging that the risk level is medium, and further evaluation is needed; when the risk prediction coefficient of the noise cavity region with the cracking risk after normalization is in a preset third threshold value interval, judging that the risk level is higher, and taking precautionary measures; When the risk prediction coefficient of the noise cavity region with the rupture risk after normalization is in a preset fourth threshold value interval, judging that the risk level is extremely high, and carrying out emergency treatment and intervention.
10. The semantic segmentation-based aortic dissection fracture risk intelligent identification method according to claim 3, further comprising: And when the normalized cavity wall morphology irregularity of the cavity wall connected domain is smaller than a preset irregularity threshold, judging that the cavity wall connected domain is a non-noise cavity region.

Description

Intelligent aortic dissection rupture risk identification method and system based on semantic segmentation Technical Field The invention relates to the technical field of image data processing, in particular to an intelligent aortic dissection fracture risk identification method and system based on semantic segmentation. Background Aortic dissection refers to the tearing of the aortic intima, and blood flows into the middle layer of the vascular wall through the rupture to form a false cavity parallel to the original true cavity. The false cavity is only wrapped by a weak adventitia, has fragile structure, not only seriously affects normal blood flow, but also can cause ischemia of important branch blood vessels. If the outer membrane of the false cavity cannot bear the pressure of blood flow to rupture, blood can flow into the chest cavity, the abdominal cavity or the pericardial cavity instantaneously, so that the patient is suddenly killed, and the complications of aortic dissection are the most critical. Therefore, the risk of the fracture is accurately identified before the fracture occurs, and the method has important significance for competing for clinical intervention time and improving the prognosis of patients. Currently, automatic identification of aortic dissection fracture risk mainly depends on semantic segmentation technology of CT images. According to the method, the structure such as a true cavity and a false cavity is extracted by classifying the CT image at the pixel level, and morphological analysis is further carried out based on a segmentation result, wherein the maximum diameter of an aortic dissection is generally used as a main static index for evaluating the fracture risk. However, the method has obvious limitations that firstly, static measurement at a single time point is relied on, an individuation baseline contrast is lacking, differences among different patients cannot be reflected, secondly, time sequence information is not introduced, dynamic evolution trend of an interlayer form is difficult to capture, thirdly, the whole diameter is taken as a criterion, and fine features of a local cavity wall form are ignored, so that false positive or false negative misjudgment easily occurs in clinical application, and the accuracy of risk identification is low. Disclosure of Invention The invention provides an intelligent aortic dissection rupture risk identification method and system based on semantic segmentation, which aim to solve the existing problems. The intelligent identification method and system for aortic dissection rupture risk based on semantic segmentation provided by the invention adopt the following technical scheme: The embodiment of the invention provides an intelligent identification method for aortic dissection fracture risk based on semantic segmentation, which comprises the following steps: Acquiring CT images of the same patient at different time points, wherein one time point corresponds to one CT image; Carrying out semantic segmentation on the CT image to obtain a plurality of connected domains of an aortic region in the CT image, wherein the plurality of connected domains at least comprise a true cavity connected domain, a false cavity connected domain and a cavity wall connected domain; acquiring the cavity wall morphology irregularity of the cavity wall communicating domain by using the Hough circle detection result of the cavity wall communicating domain, the edge pixel points and the geometric center of the cavity wall communicating domain; determining a noisy cavity region based on cavity wall morphology irregularities; acquiring a non-fracture prediction coefficient of the noise cavity region by using the true and false connected region of the noise cavity region; determining a risk trend index of the noise cavity region according to the non-fracture prediction coefficient of the noise cavity region; Determining a noise cavity region with a cracking risk according to the risk trend index of the noise cavity region; acquiring a risk prediction coefficient of the noise cavity region with the cracking risk based on the risk trend index of the noise cavity region with the cracking risk and the cavity wall morphology irregularity; and acquiring a risk level evaluation result according to the risk prediction coefficient of the noise cavity region with the rupture risk. Further, the obtaining the cavity wall morphology irregularity of the cavity wall communicating domain by using the hough circle detection result of the cavity wall communicating domain, the edge pixel point and the geometric center of the cavity wall communicating domain comprises the following specific steps: Calculating the absolute value of the difference between the abscissa of the edge pixel point and the abscissa of the geometric center of the cavity wall communicating domain and the absolute value of the difference between the ordinate of the edge pixel point and the ordinate of the ge