CN-121983243-A - Three-dimensional organ reconstruction registration fusion method based on multi-modal medical image

CN121983243ACN 121983243 ACN121983243 ACN 121983243ACN-121983243-A

Abstract

The invention provides a three-dimensional organ reconstruction registration fusion method based on multi-mode medical images, and belongs to the technical field of medical images. The method comprises the steps of S1, collecting multi-mode medical images of an ith organ of a target patient in real time, constructing a preoperative data set, S2, continuously passing through an intraoperative ultrasonic image after constructing a first data set, constructing an intraoperative data set, S3, constructing an AI model by utilizing a convolutional neural network, and S4, constructing an ith organ stability coefficient based on the preoperative data set and the intraoperative data set And crowding degree coefficient, S5, the ith organ stability coefficient And (3) constructing an integrated dynamic risk coefficient by correlating with the ith organ crowding coefficient. According to the invention, the CT, MRI and intraoperative ultrasonic images are simultaneously acquired, the preoperative data set is constructed, the change rate of the organ volume and the body position adjustment displacement are updated in real time in the operation, the optimal operation path can be automatically predicted, and subjectivity depending on doctor experience in traditional operation path planning is reduced.

Inventors

XIAO WEIDONG
QIN JIAJIA
WANG ZIHAN
LI JIAXIN
JIANG ENLAI
DU GUANGSHENG
QIU YUAN
LI LIQI
WANG WENSHENG
ZHENG DAOFENG

Assignees

中国人民解放军陆军军医大学第二附属医院

Dates

Publication Date: 20260505
Application Date: 20251204

Claims (10)

1. A three-dimensional organ reconstruction registration fusion method based on multi-modal medical images is characterized by comprising the following steps: S1, acquiring multi-mode medical images of an ith organ of a target patient in real time, including CT images and MRI images, and extracting the volume of the ith organ Three-dimensional space coordinates Distance adjacent to the (i+1) th organ Local displacement vector Constructing a preoperative dataset; s2, after the first data set is constructed, continuously acquiring dynamic and morphological data of the ith organ in real time through intraoperative ultrasonic images, wherein the dynamic and morphological data comprise local volume change rate of the ith organ caused by intraoperative operation Relative displacement of the ith organ caused by intra-operative posture adjustment Constructing an intraoperative dataset; S3, constructing an AI model by using a convolutional neural network, inputting a preoperative dataset and an intraoperative dataset into the AI model, and predicting and outputting an operation path; s4, constructing an ith organ stability coefficient based on the preoperative data set and the intraoperative data set And a congestion degree coefficient And the ith organ stability factor In contrast to stability threshold A, when the ith organ stability factor When the stability threshold value A is smaller than or equal to the stability threshold value A, a first alarm instruction is generated, and the ith organ crowding degree coefficient is obtained When the ith organ crowding factor is compared with the crowding threshold value Q Generating a second alarm instruction when the congestion degree is larger than the congestion degree threshold value Q; s5, the ith organ stability coefficient And the ith organ crowding factor Correlating, constructing comprehensive dynamic risk coefficient And will integrate dynamic risk factors Compared with the comprehensive risk threshold W, when the comprehensive dynamic risk coefficient And if the total risk is greater than the comprehensive risk threshold W, evaluating and generating an optimization instruction.
2. The method for registration fusion of three-dimensional organ reconstruction based on multi-modal medical image according to claim 1, wherein the step S1 comprises: S11, acquiring preoperative CT image and MRI image data by shooting a CT image and an MRI image of a target patient, establishing a three-dimensional coordinate system x, y and z, performing three-dimensional segmentation on the CT image and the MRI image, separating an ith organ from the CT image and the MRI image by adopting edge detection on the CT image and the MRI image, acquiring a three-dimensional segmentation area of the ith organ, and calculating the volume of the ith organ based on the number of voxels of the segmentation area and the physical volume of the voxels ; S12, calculating the mass center of the target organ in the established three-dimensional coordinate system based on the three-dimensional segmentation areas of the preoperative CT image and the MRI image to obtain the three-dimensional space coordinate of the ith organ ; S13, respectively carrying out three-dimensional segmentation on the ith organ and the (i+1) th organ in the CT image and the MRI image, extracting a surface point set, and obtaining the adjacent distance between the ith organ and the (i+1) th organ by calculating the shortest Euclidean distance between the surface point set of the target organ and the surface point set of the adjacent organ ; S14, respectively shooting a CT image and an MRI image through an nth time point and an n+1th time point, rigidly registering the CT image and the MRI image shot at the nth time point with the CT image and the MRI image shot at the n+1th time point, calculating a space three-dimensional coordinate difference value of the ith organ at the nth time point and the n+1th time point, and obtaining a local displacement vector of the ith organ Thereby constructing a preoperative dataset.
3. The method for registration fusion of three-dimensional organ reconstruction based on multi-modal medical image according to claim 2, wherein the step S2 comprises: S21, continuously acquiring an ith organ image of an intraoperative target patient through an intraoperative color Doppler ultrasonic machine, performing three-dimensional segmentation on the ith organ image of each frame, counting the number of segmented voxels, and calculating to obtain the local volume change rate of the ith organ caused by intraoperative operation ; S22, adjusting the body position of the target patient in the operation, acquiring an image of an ith organ in real time by adopting a color Doppler ultrasonic machine, extracting three-dimensional space coordinates of the ith organ in the image based on a three-dimensional coordinate system, registering the three-dimensional space coordinates with the three-dimensional space coordinates of the ith organ before the operation, calculating local coordinate difference values of the ith organ before and after the body position adjustment, and obtaining the relative displacement of the ith organ caused by the body position adjustment in the operation 。
4. The method for registration fusion of three-dimensional organ reconstruction based on multi-modal medical image according to claim 3, wherein the step S3 comprises: S31, constructing an AI model by using a convolutional neural network, training and testing the AI model by using a preoperative data set and an intraoperative data set, taking the trained AI model as a three-dimensional organ reconstruction registration fusion evaluation model of the multi-mode medical image, simultaneously using the output of the equipment operation AI model as a feature vector to identify feature information, and taking the trained AI model as data operation prediction.
5. The method for registration fusion of three-dimensional organ reconstruction based on multi-modal medical image according to claim 4, wherein the step S4 comprises: S41, based on the volume of the ith organ in the preoperative dataset Three-dimensional coordinates Distance adjacent to the (i+1) th organ Local displacement vector The ith organ stability factor is obtained by calculation in the following manner; First, for the volume of the ith organ Three-dimensional coordinates Distance adjacent to the (i+1) th organ Local displacement vector Normalizing to obtain the volume of the ith organ after normalization Three-dimensional coordinates Distance adjacent to the (i+1) th organ Local displacement vector ; In the formula (I), in the formula (II), Expressed as the minimum volume of the organ, Expressed as the maximum volume of the organ; Three-dimensional coordinates for the ith organ Performing normalization processing, and calculating deviation vector of three-dimensional coordinates of the ith organ ; In the formula (I), in the formula (II), Represented as the three-dimensional spatial coordinates of the ith organ, A reference three-dimensional spatial coordinate represented as an ith organ; the deviation module length is then calculated ; In the formula (I), in the formula (II), Represented as the x-axis coordinate value of the three-dimensional space coordinate of the ith organ, A reference coordinate value expressed as the x-axis of the three-dimensional space coordinate of the ith organ, Expressed as y-axis coordinate values of three-dimensional space coordinates of the ith organ, A reference coordinate value expressed as a y-axis of three-dimensional space coordinates of the ith organ, Represented as z-axis coordinate values of three-dimensional space coordinates of the ith organ, A reference coordinate value expressed as a z-axis of three-dimensional space coordinates of the ith organ; finally, for the three-dimensional coordinates of the ith organ Carrying out normalization treatment; In the formula (I), in the formula (II), Expressed as the module length of the deviation between the current coordinates of the ith organ and the reference coordinates, Expressed as the maximum deviation module length of the organ; In the formula (I), in the formula (II), Expressed as the minimum distance between adjacent organs, Expressed as the maximum distance between adjacent organs; Local displacement vector for the ith organ Carrying out normalization treatment; first, the local displacement vector modular length of the ith organ is calculated ; In the formula (I), in the formula (II), Represented as a local displacement component of the ith organ in the x-axis direction, Represented as a local displacement component of the ith organ in the y-axis direction, Expressed as a local displacement component of the ith organ in the z-axis direction; Then to the local displacement vector of the ith organ Carrying out normalization treatment to obtain; In the formula (I), in the formula (II), Expressed as the maximum value of the modulus of the local displacement vector of the organ; according to the volume of the ith organ Three-dimensional coordinates Distance adjacent to the (i+1) th organ Local displacement vector Obtaining the stability coefficient of the ith organ through formula calculation 。
6. The method for registration fusion of three-dimensional organ reconstruction based on multi-modal medical image according to claim 5, wherein the step S4 further comprises: S42, the ith organ stability factor Combining the historical data, analyzing the value of the stability threshold A from the historical data, and combining experience to judge the stability threshold A; When (when) When > A, the position and the morphology of the ith organ in the target patient are normal in operation and are used as references for planning an operation path; When (when) And when the value is less than or equal to A, representing that the position and the morphology of the ith organ in the target patient are abnormal in operation, generating a first alarm instruction, and adopting an optimization strategy, wherein the optimization strategy comprises the steps of adjusting the incision angle of an operation path based on the offset direction and the amplitude of the ith organ, correcting the incision direction according to 10-30% of the offset, and improving the image acquisition frequency of the ith organ by 20-70%.
7. The method for registration fusion of three-dimensional organ reconstruction based on multi-modal medical image according to claim 6, wherein the step S4 further comprises: s43, local volume change rate of the ith organ caused by operation in the data set in operation Relative displacement of the ith organ caused by intra-operative posture adjustment The ith organ crowding factor is obtained by ; First to the local volume change rate of the ith organ caused by the operation Relative displacement of the ith organ caused by intra-operative posture adjustment Performing normalization treatment to obtain local volume change rate of ith organ caused by normalized operation Relative displacement of the ith organ caused by intra-operative posture adjustment ; In the formula (I), in the formula (II), Expressed as the maximum value of the rate of change of the volume of the organ; In the formula (I), in the formula (II), Expressed as organ maximum displacement value; based on the local volume change rate of the ith organ caused by the intraoperative procedure Relative displacement of the ith organ caused by intra-operative posture adjustment Obtaining the ith organ crowding degree coefficient through formula calculation 。
8. The method for registration fusion of three-dimensional organ reconstruction based on multi-modal medical image according to claim 7, wherein the step S4 further comprises: S44, the ith organ crowding degree coefficient Comparing with a crowding degree threshold Q; When (when) When Q is greater than the value, the volume change rate and the relative displacement of the ith organ of the target patient are abnormal, a second alarm instruction is generated, and a correction strategy is adopted, wherein the correction strategy comprises the steps of correcting the incidence angle of the surgical path by 12% -33% based on the relative displacement direction of the ith organ, adjusting the traction force and the negative pressure suction by 5% -20%, and adding 5% -15% of spatial offset compensation to the surgical path; When (when) And when Q is less than or equal to Q, the volume change rate and the relative displacement of the ith organ of the target patient are normal.
9. The method for registration fusion of three-dimensional organ reconstruction based on multi-modal medical image according to claim 8, wherein the step S5 comprises: s51, the ith organ stability factor And the ith organ crowding factor In association, the comprehensive dynamic risk coefficient is obtained by ; 。
10. The method for registration fusion of three-dimensional organ reconstruction based on multi-modal medical image according to claim 9, wherein the step S5 further comprises: s52, integrating dynamic risk coefficients Comparing with a comprehensive risk threshold W; When (when) When W, representing that the position of the ith organ of the target patient is abnormal in the operation, generating an optimization instruction, wherein the optimization instruction comprises correcting the cutting angle of an operation path according to the displacement direction of the ith organ by 10% -30%, and expanding a safety boundary by 5% -15% based on the change of the crowding degree; When (when) And when W is less than or equal to W, indicating that the ith organ of the target patient is positioned normally in the operation.

Description

Three-dimensional organ reconstruction registration fusion method based on multi-modal medical image Technical Field The invention relates to the technical field of medical images, in particular to a three-dimensional organ reconstruction registration fusion method based on multi-mode medical images. Background Along with the rapid development of minimally invasive surgery, accurate grasp of the position, form and spatial relationship of organs of a patient in surgery becomes a key factor for improving the safety and accuracy of surgery, in complex surgery areas such as abdomen and the like, the structure between different organs is compact, dynamic change is obvious, and the organs are influenced by various factors such as respiratory motion, organ traction, negative pressure suction, surgical instrument intervention, patient position change and the like, and real-time volume change, spatial displacement and local deformation can occur to the organs, so that a surgery path formulated before surgery can deviate from the actual organ state, the operation risk is increased, for example, in liver and gall surgery and pancreas surgery, the small displacement of adjacent organs can lead to insufficient safety distance between a surgical instrument and a target structure, and bleeding and accidental injury of adjacent organs in surgery can be caused. At present, three-dimensional models are generated by commonly relying on static images such as CT (computed tomography), MRI (magnetic resonance imaging) and the like before operation and used for planning before operation and making a path, however, the images only reflect the static state of a patient at a certain moment, the dynamic behavior of the organ in operation along with time change cannot be represented, and the dynamic behavior is difficult to keep synchronous with the rapid change of the organ in an actual operation scene, so that a three-dimensional organ reconstruction registration fusion method based on multi-mode medical images is provided for solving the problem. Disclosure of Invention In order to overcome the above disadvantages, the present invention provides a three-dimensional organ reconstruction registration fusion method based on multi-modal medical images, which overcomes or at least partially solves the above technical problems. The invention is realized in the following way: The invention provides a three-dimensional organ reconstruction registration fusion method based on multi-mode medical images, which comprises the following steps: S1, acquiring multi-mode medical images of an ith organ of a target patient in real time, including CT images and MRI images, and extracting the volume of the ith organ Three-dimensional space coordinatesDistance adjacent to the (i+1) th organLocal displacement vectorConstructing a preoperative dataset; s2, after the first data set is constructed, continuously acquiring dynamic and morphological data of the ith organ in real time through intraoperative ultrasonic images, wherein the dynamic and morphological data comprise local volume change rate of the ith organ caused by intraoperative operation Relative displacement of the ith organ caused by intra-operative posture adjustmentConstructing an intraoperative dataset; S3, constructing an AI model by using a convolutional neural network, inputting a preoperative dataset and an intraoperative dataset into the AI model, and predicting and outputting an operation path; s4, constructing an ith organ stability coefficient based on the preoperative data set and the intraoperative data set And a congestion degree coefficientAnd the ith organ stability factorIn contrast to stability threshold A, when the ith organ stability factorWhen the stability threshold value A is smaller than or equal to the stability threshold value A, a first alarm instruction is generated, and the ith organ crowding degree coefficient is obtainedWhen the ith organ crowding factor is compared with the crowding threshold value QGenerating a second alarm instruction when the congestion degree is larger than the congestion degree threshold value Q; s5, the ith organ stability coefficient And the ith organ crowding factorCorrelating, constructing comprehensive dynamic risk coefficientAnd will integrate dynamic risk factorsCompared with the comprehensive risk threshold W, when the comprehensive dynamic risk coefficientAnd if the total risk is greater than the comprehensive risk threshold W, evaluating and generating an optimization instruction. In a preferred embodiment, the step S1 includes: S11, acquiring preoperative CT image and MRI image data by shooting a CT image and an MRI image of a target patient, establishing a three-dimensional coordinate system x, y and z, performing three-dimensional segmentation on the CT image and the MRI image, separating an ith organ from the CT image and the MRI image by adopting edge detection on the CT image and the MRI image, acquiring a three-dimensional s