CN111973271B

CN111973271B - Preoperative ablation region simulation method and device for tumor thermal ablation

Info

Publication number: CN111973271B
Application number: CN202010894336.2A
Authority: CN
Inventors: 付天宇; 杨健; 宋红; 米龙哲; 范敬凡
Original assignee: Beijing Institute of Technology BIT
Current assignee: Beijing Institute of Technology BIT
Priority date: 2020-08-31
Filing date: 2020-08-31
Publication date: 2021-12-14
Anticipated expiration: 2040-08-31
Also published as: CN111973271A

Abstract

A preoperative ablation region simulation method and device for tumor thermal ablation are provided, the method comprises the following steps: (1) selecting data suitable for training, and marking preoperative and postoperative images by a doctor to obtain preoperative livers, postoperative ablation regions and needle tracks; (2) registering the postoperative image to the preoperative image, adding rigid limitation in the registration process to ensure that the ablation region and the needle track region do not generate elastic deformation, and projecting the needle track information and the ablation region onto the preoperative image by utilizing a deformation field obtained by registration; (3) generating an ideal thermal field distribution map according to the projected needle track direction and position information and the corresponding power time adopted in the operation; (4) and establishing a convolutional neural network, cutting the ideal thermal field distribution map and the preoperative image according to a preoperative liver mask, combining cutting results as network input, learning by taking the deformed ablation region as a gold standard to obtain a regression model, and realizing the prediction of the ablation region in a test stage.

Description

Preoperative ablation region simulation method and device for tumor thermal ablation

Technical Field

The invention relates to the technical field of medical image processing, in particular to a preoperative ablation region simulation method for tumor thermal ablation and a preoperative ablation region simulation device for tumor thermal ablation, and the method and the device are mainly suitable for an ablation operation based on deep learning guidance.

Background

During the microwave ablation operation, a doctor puts an ablation needle into a tissue focus position through percutaneous puncture, microwaves are emitted from the front electrode of the ablation needle, the temperature of surrounding focus tissues rises after the surrounding focus tissues are radiated by the microwaves, and when the thermal damage of the surrounding focus tissues exceeds a certain limit, the proteins of the surrounding focus tissues are inactivated to generate an ablation area, so that the treatment of tumors is realized. In order to provide the optimal ablation method corresponding to different patients for doctors and realize the individualized conformal ablation, the position and the direction of a needle are designed according to the position of a tumor and important tissue blood vessels provided by preoperative images in the preoperative planning stage, and on the basis, the range of an ablation region is simulated and generated. The position and the direction of the needle are adjusted by comparing the position size relationship between the ablation area and the tumor, so that the optimal ablation scheme suitable for the patient is finally determined, the real ablation in the follow-up operation is guided, and the overall operation efficiency is improved. Therefore, in preoperative planning, the prediction of the size and shape of the ablation region is the basis for designing the optimal ablation scheme according to the swing position direction of the ablation needle.

The cause of the ablation area is complex, and the size and the shape of the ablation area are not only related to the power time of an ablation needle, but also influenced by complex factors such as the heat sink effect of surrounding blood vessels, the tissue tumor tissue characteristics and the like. Currently, the ablation region generation models used are classified into an ellipsoid model and a tissue-based Specific Absorption Rate (SAR) distribution model. For the first model, the ablation region is approximated as an ellipsoid, the major axis of the ellipsoid being the direction of the needle insertion of the ablation needle, and the minor axis being user-defined or set to an empirical value. The SAR distribution model reflects the temperature rise of the tissues after absorbing the microwaves, and the range of an ablation region generated by the ablation needle can be determined by calculating the thermal injury through an Arrhenius equation. The SAR distribution model and the Arrhenius thermal injury equation comprise the parameters of density and specific heat of tissues and blood, tissue thermal conductivity, blood perfusion rate and the like. Thus, compared to an ellipsoid model, the SAR distribution model contains important factors that influence the generation of the ablation region. However, the SAR model requires an accurate blood vessel segmentation result when calculating the vascular heat sink effect, and meanwhile, because the related parameters of different patient tissues and blood are different and the actual values of the parameters are difficult to obtain, the ablation region solved by the SAR distribution model still has a large difference from the actual ablation effect. This seriously reduces the clinical guidance significance of preoperative planning on subsequent real operations.

Disclosure of Invention

In order to overcome the defects of the prior art, the invention provides a preoperative ablation region simulation method aiming at tumor thermal ablation, which not only avoids the acquisition of a large number of complex parameters and the segmentation of relevant tissues such as blood vessels, but also predicts that the obtained ablation region is closer to the real ablation effect, so that preoperative planning has clinical guidance significance for subsequent real operations.

The technical scheme of the invention is as follows: the preoperative ablation region simulation method for tumor thermal ablation comprises the following steps:

(1) selecting data suitable for training, and marking preoperative and postoperative images by a doctor to obtain preoperative livers, postoperative ablation regions and needle tracks;

(2) registering the postoperative image to the preoperative image, adding rigid limitation in the registration process to ensure that the ablation region and the needle track region do not generate elastic deformation, and projecting the needle track information and the ablation region onto the preoperative image by utilizing a deformation field obtained by registration;

(3) generating an ideal thermal field distribution map according to the projected needle track direction and position information and the corresponding power time adopted in the operation;

(4) establishing a convolutional neural network, cutting the ideal thermal field distribution diagram and the preoperative image according to a preoperative liver mask, combining cutting results as network input, learning by taking the deformed ablation region as a gold standard to obtain a regression model, and testing the regression model

The ablation zone is now predicted.

The retrospective ablation data are utilized, a regression model is built through deep learning on the basis of a real ablation result, and an ablation area is predicted. Different from an ellipsoid and SAR distribution model prediction method, the invention takes the real ablation area obtained after the operation as a gold standard, takes the image before the operation and the ideal thermal field generated by an ablation needle as input, and utilizes a convolutional neural network to establish a regression model, thereby not only avoiding the acquisition of a large number of complex parameters and the segmentation of relevant tissues such as blood vessels, but also taking the real ablation area as the gold standard, and the predicted ablation area is more approximate to the real ablation effect, so that the planning before the operation has clinical guidance significance for the subsequent real operation.

Also provided is a preoperative ablation zone simulation device for tumor thermal ablation, comprising:

a labeling module configured to select data suitable for training, and to label preoperative and postoperative images by a doctor to obtain preoperative livers, postoperative ablation regions and needle tracks;

the registration module is configured to register the postoperative image to the preoperative image, add rigidity limitation in the registration process to ensure that the ablation region and the needle channel region do not generate elastic deformation, and project the needle channel information and the ablation region onto the preoperative image by utilizing a deformation field obtained by registration;

a thermal field map generation module configured to generate an ideal thermal field map based on projected needle track direction and position information and corresponding intraoperative power time;

the deep learning module is configured to establish a convolutional neural network, cut the ideal thermal field distribution diagram and the preoperative image according to a preoperative liver mask, combine the cutting results as network input, learn by taking the deformed ablation region as a gold standard to obtain a regression model, and realize prediction of the ablation region in a testing stage.

Drawings

Fig. 1 is a flow chart of a method of preoperative ablation zone simulation for tumor thermal ablation according to the present invention.

Detailed Description

As shown in fig. 1, the method for simulating a preoperative ablation region for tumor thermal ablation comprises the following steps:

(4) and establishing a convolutional neural network, cutting the ideal thermal field distribution map and the preoperative image according to a preoperative liver mask, combining cutting results as network input, learning by taking the deformed ablation region as a gold standard to obtain a regression model, and realizing the prediction of the ablation region in a test stage.

The ablation data for each patient includes preoperative and postoperative MR images as well as power and ablation time information of an ablation needle used in the operation. Preferably, in step (1), in order to enable the network to learn the influence of the vascular heat sink effect, images of labeling, subsequent registration and prediction training performed on preoperative and postoperative images by a doctor are in a venous phase, and the contrast between hepatic vessels in the venous phase and the liver is stronger than that in other phases; drawing outlines of the liver and the ablation region when marking the liver and the ablation region; the marking of the needle path is that a needle point of the ablation needle and an intersection point world coordinate of the ablation needle and an ablation area are marked on the postoperative image, and the ablation needle is regarded as a ray passing through two marked points in the postoperative image because the ablation needle is approximately a rigid body, wherein the top point of the ray is the needle point.

Preferably, in the step (1), after the labeling is completed, mask images of the preoperative liver L and the postoperative ablation region a, and a placing position and a direction parameter of an ablation needle are respectively obtained.

The poses of the patients in the images before and after the operation may be different, meanwhile, the liver is an elastic tissue, and due to the respiratory influence, the liver in the images before and after the operation is subjected to non-rigid deformation, so that the images before and after the operation are matched in a two-step registration mode. Preferably, the step (2) comprises the following substeps:

(2.1) carrying out rigid registration on the postoperative image to the preoperative image, eliminating the difference of translation and rotation generated by the posture of the patient through rigid registration, and obtaining a deformation matrix after rigid registration;

and (2.2) solving the deformation field in the liver region by adopting a non-rigid registration algorithm to further reduce the difference between the preoperative and postoperative images.

Preferably, in the step (2.2), in order to avoid deformation of the ablation region and the needle track, the ablation region and the needle track are kept rigid during the registration process; for the ablation region, its rigid body is maintained by the rigid constraint term during the registration process; for an ablation needle, since the marker points are all within the ablation zone, the rigid limitation of the ablation zone also keeps the ablation needle rigid during registration.

Preferably, in the step (3), the thermal field generated by the ideal SAR model is distributed in an axisymmetric manner with the ablation needle as an axis, and a thermal field image is obtained according to the deformed ablation needle information and the ideal SAR distribution model.

Preferably, in the step (4), the prediction of the ablation region is realized by using a U-Net network, the input data is a merged image of the MR image and the thermal field image before the operation, the deformed ablation region mask image is used as a gold standard, the image at the vein stage before the operation provides important blood vessel and liver tumor characteristic information, and the thermal field image provides important ablation needle characteristic information.

Preferably, in the step (4), the preoperative image and the thermal field image are cropped by using the labeled liver mask, and the predicted overlap ratio between the ablation region and the real ablation region is used as the loss function loss. Through learning, an ablation region prediction model is finally obtained, and rapid and accurate prediction of an ablation region can be realized for input test data.

It will be understood by those skilled in the art that all or part of the steps in the method of the above embodiments may be implemented by hardware instructions related to a program, the program may be stored in a computer-readable storage medium, and when executed, the program includes the steps of the method of the above embodiments, and the storage medium may be: ROM/RAM, magnetic disks, optical disks, memory cards, and the like. Therefore, corresponding to the method of the present invention, the present invention also includes a preoperative ablation zone simulation device for tumor thermal ablation, which is generally represented in the form of functional modules corresponding to the steps of the method. The device includes:

The above description is only a preferred embodiment of the present invention, and is not intended to limit the present invention in any way, and all simple modifications, equivalent variations and modifications made to the above embodiment according to the technical spirit of the present invention still belong to the protection scope of the technical solution of the present invention.

Claims

1. A preoperative ablation region simulation method aiming at tumor thermal ablation is characterized by comprising the following steps: which comprises the following steps:

2. The method of preoperative ablation zone simulation for tumor thermal ablation according to claim 1, characterized by: in the step (1), images which are labeled by a doctor before and after operation and are subjected to subsequent registration and prediction training are in a venous phase, and the contrast between hepatic vessels in the venous phase and the liver is stronger than that between other phases; drawing outlines of the liver and the ablation region when marking the liver and the ablation region; the marking of the needle path is that a needle point of the ablation needle and an intersection point world coordinate of the ablation needle and an ablation area are marked on the postoperative image, and the ablation needle is regarded as a ray passing through two marked points in the postoperative image because the ablation needle is approximately a rigid body, wherein the top point of the ray is the needle point.

3. The method of preoperative ablation zone simulation for tumor thermal ablation according to claim 2, characterized by: in the step (1), after the marking is finished, a preoperative liver L and a postoperative ablation region A mask image, and an ablation needle placing position and direction parameters are respectively obtained.

4. The method of simulating a preoperative ablation zone for tumor thermal ablation according to claim 3, characterized in that: the step (2) comprises the following sub-steps:

5. The method of preoperative ablation zone simulation for tumor thermal ablation according to claim 4, characterized by: in the step (2.2), in order to avoid deformation of the ablation region and the needle path, the ablation region and the needle path are kept as rigid bodies in the registration process; for the ablation region, its rigid body is maintained by the rigid constraint term during the registration process; for an ablation needle, since the marker points are all within the ablation zone, the rigid limitation of the ablation zone also keeps the ablation needle rigid during registration.

6. The method of preoperative ablation zone simulation for tumor thermal ablation according to claim 5, characterized by: in the step (3), the thermal field generated by the ideal SAR model is distributed in an axisymmetric manner by taking the ablation needle as an axis, and a thermal field image is obtained according to the information of the deformed ablation needle and the ideal SAR distribution model.

7. The method of preoperative ablation zone simulation for tumor thermal ablation according to claim 6, characterized by: in the step (4), the prediction of the ablation region is realized by adopting a U-Net network, the input data is a merged image of an MR image and a thermal field image before an operation, the deformed ablation region mask image is used as a gold standard, the image of the vein stage before the operation provides important blood vessel and liver tumor characteristic information, and the thermal field image provides important ablation needle characteristic information.

8. The method of preoperative ablation zone simulation for tumor thermal ablation according to claim 7, wherein: in the step (4), the labeled liver mask is used to crop the preoperative image and the thermal field image, and the predicted overlapping rate between the ablation region and the real ablation region is used as the loss function loss.

9. Melt regional analogue means before art to tumour heat ablation, its characterized in that: it includes: a labeling module configured to select data suitable for training, and to label preoperative and postoperative images by a doctor to obtain preoperative livers, postoperative ablation regions and needle tracks;