A method and system for guiding the planning of ablation of renal lesions

Abstract

The present invention discloses a method and system for guiding the planning of ablation of a renal lesion, which relates to the technical field of ablation guidance planning. The method and system for guiding the planning of ablation of a renal lesion include the following specific steps: acquiring the anatomical structure of the kidney based on image acquisition, extracting key features from the acquired multi-source images, extracting and identifying the edge of the tumor, identifying the edge of the tumor based on the acquired key features, and generating an optimal ablation path that avoids the key structure. The present invention acquires the anatomical structure of the kidney through multiple CT and ICE images to reduce the error in identifying the boundary of the tumor, and aligns the multi-source images based on a deformation field algorithm to further improve the accuracy of identifying the boundary of the tumor. In the actual ablation process, the multi-source images are fused with an electromagnetic sensor to monitor the deviation of the ablation needle tip.

Description

Technical Field

[0001] The present invention relates to the technical field of ablation guidance planning, and in particular to a method and system for ablation guidance planning of a renal lesion. Background Art

[0002] Thermal ablation involves destroying tissue through extreme heat and is a minimally invasive alternative to resection and transplantation for the treatment of liver tumors. Thermal ablation for liver cancer has become a first-line curative treatment for the tumor because it has similar overall survival rates to surgical resection but is much less invasive, has lower complication rates, is cost-effective, and has a very low treatment-related mortality rate.

[0003] Renal lesion ablation is a minimally invasive treatment method mainly used to treat kidney tumors, including benign and malignant tumors. Under image guidance, radiofrequency ablation inserts a radiofrequency needle into the tumor, and discharges heat through the needle tip, causing coagulative necrosis of tumor cells. Cryoablation uses low temperature to destroy tumor cells. Usually, a freezing probe is inserted into the tumor under image guidance, and cell death is caused by rapid cooling. Microwave ablation uses microwave radiation to heat the tumor tissue to a lethal temperature, thereby killing tumor cells.

[0004] However, the tumor boundary relies on experience-based judgment, and the ablation tip in traditional ablation is unstable, resulting in uneven energy transfer. In addition, due to reliance on experience, the ablation tip may accidentally touch normal areas during the ablation process, causing necrosis or freezing of normal tissue cells. In addition, during ablation planning, there is a mixture of multiple ablation methods, making it impossible to change the ablation measures in real time for complex lesions. Summary of the Invention

[0005] In view of the deficiencies in the prior art, the present invention provides a method and system for guiding the planning of ablation of a renal lesion, which solves the problems raised in the above-mentioned background technology.

[0006] To achieve the above objectives, the present invention is implemented through the following technical solutions: a method and system for guiding the planning of ablation of a renal lesion, the method comprising the following specific steps:

[0007] S100, based on image acquisition, acquires renal anatomical structure, marks vascular branches and variations, and captures vascular wall micromotion. It also aligns the acquired multi-source images based on a learning algorithm to extract key features.

[0008] S200, completing tumor edge recognition based on the acquired key features and generating an optimal ablation path that avoids key structures;

[0009] S300, with electromagnetic sensor fusion multi-source image, monitors the deviation of the ablation needle tip and introduces the impedance change rate to dynamically adjust the radiofrequency power;

[0010] S400. After surgery, integrate clinical data and pathway points and store them in the clinical database.

[0011] A further improvement of the technical solution of the present invention is that in said S100, the image acquisition based on which the anatomical structure of the kidney is acquired, the vascular branches and variations are marked, and the micro-motion of the vascular wall is captured is , the multi-source images aligned based on the learning algorithm are , the S100 is inputted through the deformation field Finish Every spatial point The displacement vector to The transformation and The transformation relationship between them is, , where is the deformation field, representing each spatial point The displacement vector, is the regularization coefficient, which is used to control the smoothness of the deformation field. The acquisition of renal anatomical structure is based on medical CT, the marking of vascular branches and variations is based on DSA, and the capture of vascular wall micro-motion is based on ICE. After alignment, they are respectively recorded as 、 、 , input into the fusion model to complete the key feature extraction.

[0012] A further improvement of the technical solution of the present invention is that the fusion model in S100 includes: ,in 、 as well as are the weight coefficients of CT, ICE and DSA respectively. The weight coefficients are generated by the attention mechanism, specifically: , where is the input aligned image data, is the global average pooling, For sigmid( ) function, output weight value 0-1, MLP is a multi-layer perceptron, after obtaining the fused image Then, the kidney tumor area is segmented to distinguish normal tissue from the lesion area, and the lesion area is used as the key feature, and each point in the lesion area is labeled, recorded as In addition, after obtaining the fused image After that, perform basic image segmentation to obtain the number of vascular skeleton branches after segmentation. and the overall volume , according to the mathematical formula Obtain vascular branch density and determine vascular wall displacement based on grayscale values , which is convenient for judging the subsequent ablation path. is the grayscale change in the time dimension, is the spatial gradient, To prevent the zero constant, the vascular branching density and vascular wall displacement are conventional features in addition to the key features.

[0013] A further improvement of the technical solution of the present invention is that in S200, the key structures avoided by the optimal ablation path include blood vessels and tumor areas, and the planning of the optimal ablation path includes the following specific steps:

[0014] The lesion area obtained in S100 is divided into a safe area and a dangerous area, and the doctor determines the starting point of the ablation path. , based on the A-Star path planning model, the cost function of the ablation path is established, specifically: , where From the starting point to the current point The actual cost, is the penalty weight, The risk penalty term is introduced because the distance between the multiple points P on the path and the neighboring point k in the safe zone is too close. Therefore, during ablation, it is easy to touch the tissue or organ in the safe zone. , For waypoints The Euclidean distance to the neighboring point k, is the decay function, is the attenuation coefficient, The set of all points in the labeled lesion area is marked as belonging to the safe area or the dangerous area. For adjacent points;

[0015] In the cost function After confirmation, initialize the priority queue, starting point Join the queue, the cost = 0, iteratively expands adjacent points based on the A-Star path planning model, and calculates And update the path, when the end point Terminate when visited, and backtrack to generate the path.

[0016] The purpose of introducing the risk penalty term is to There are dangerous points in the neighborhood of ,but ,Right now Increase , high penalties force the paths to avoid the danger zone during iterations.

[0017] The further improvement of the technical solution of the present invention is that: in the above S300, the electromagnetic navigation coordinate system generated by the electromagnetic sensor is E, and the fusion image obtained in the above S100 is fused , specifically: calculate each positioning point collected by the electromagnetic sensor and Corresponding to the cost of each point coordinate, the cost sum is rotated and translated, and the smoothness is constrained to perform regularization terms to complete the offset monitoring of the ablation needle tip. Each positioning point collected by the electromagnetic sensor is the offset point of the tube head end.

[0018] A further improvement of the technical solution of the present invention is that: when calculating each positioning point collected by the electromagnetic sensor and Corresponding to the cost of each point coordinate, the cost and the cost are rotated and translated, including:

[0019] , where The first Anchor points, For each point coordinate, is the regularization coefficient, is the regularization term, The positioning points collected by the electromagnetic sensor and The cost of the corresponding point coordinates, and in rotation and translation, the positioning point and The closest distance between the corresponding points is recorded as the closest distance from the ablation needle tip to the tumor boundary. , Corresponding points ,After optimization, T includes rotation compensation of 0.5° and translation of (0.2, 0.1, -0, 1) mm to achieve registration and prevent offset.

[0020] A further improvement of the technical solution of the present invention is that the RF power is dynamically adjusted, specifically comprising the following steps:

[0021] according to Calculate the difference between each positioning point collected by the magnetic sensor and The cost and coordinates of the corresponding points are calculated by Obtain the comprehensive error of electromagnetic navigation positioning equipment , combined with the closest distance from the ablation needle tip to the tumor boundary , the introduced impedance change rate , build a control model, specifically: , where is the instrument positioning error, The shortest distance from the ablation needle tip to the tumor boundary. is the safety distance threshold, 、 、 is the weight coefficient, which is calibrated by clinical data.

[0022] like =1.2mm, =3mm, =5Ω / min, =5mm, then =0.3×sigmoid(-2×1.2)+0.5×tanh(3 / 5)+0.2×5≈-0.1W, so the power of the ablation needle tip is automatically reduced by 0.1W to avoid damaging the blood vessels.

[0023] The present invention further provides a renal lesion ablation guidance planning system, comprising:

[0024] Multimodal image fusion module; used to integrate CT, DSA, and ICE image data, complete three-dimensional spatial alignment of images, and extract vascular branch density, vascular wall displacement, and tumor boundaries;

[0025] Path planning module: constructs an ablation path that avoids key structures including blood vessels and tumor areas, and makes real-time path corrections;

[0026] Ablation navigation module; based on the electromagnetic sensor array and impedance detection, it dynamically calculates the device error and regulates the RF power output in combination with the impedance change rate.

[0027] Compared with the existing technology, the beneficial effects of the present invention are: obtaining the renal anatomical structure through CT, DSA and ICE multiple images, reducing the boundary recognition error of the tumor, and aligning the multi-source images based on the deformation field algorithm to further improve the recognition accuracy of the tumor boundary, and in the actual ablation process, the electromagnetic sensor is used to fuse the multi-source images to monitor the offset of the ablation needle tip to avoid damage to normal tissue. The introduced impedance change rate dynamically completes the radio frequency frequency of the ablation end during the ablation process, and timely ablation corresponding replacement and power adjustment are carried out for complex lesions. The data and path points involved in the process are used as optimization data for the learning model to provide a training basis for subsequent operations. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] Figure 1 Schematic diagram of the flow of the guided planning method of the present invention;

[0029] Figure 2 This is a system block diagram of the present invention. DETAILED DESCRIPTION

[0030] Various exemplary embodiments, features, and aspects of the present application will be described in detail below with reference to the accompanying drawings. The same reference numerals in the accompanying drawings represent elements with the same or similar functions. Although various aspects of the embodiments are shown in the accompanying drawings, the drawings are not necessarily drawn to scale unless otherwise indicated.

[0031] The word “exemplary” is used exclusively herein to mean “serving as an example, example, or illustration.” Any embodiment described herein as “exemplary” is not necessarily to be construed as preferred or advantageous over other embodiments.

[0032] In addition, numerous specific details are provided in the following specific examples to better illustrate the present application. Those skilled in the art will appreciate that the present application can be practiced without certain specific details. In some instances, methods, means, and components well known to those skilled in the art are not described in detail in order to highlight the main purpose of the present application.

[0033] The present invention provides a method for guiding the planning of ablation of a renal lesion, comprising the following specific steps:

[0034] S100, based on image acquisition, acquires renal anatomical structure, marks vascular branches and variations, and captures vascular wall micromotion. It also aligns the acquired multi-source images based on a learning algorithm to extract key features.

[0035] In S100, the collected images are used to obtain the anatomy of the kidney, mark the vascular branches and variations, and capture the micro-motion of the vascular wall. , the multi-source images aligned based on the learning algorithm are , the S100 is inputted through the deformation field Finish Every spatial point The displacement vector to The transformation and The transformation relationship between them is, , where is the deformation field, representing each spatial point The displacement vector, is the regularization coefficient, which is used to control the smoothness of the deformation field. The acquisition of renal anatomical structure is based on medical CT, the marking of vascular branches and variations is based on DSA, and the capture of vascular wall micro-motion is based on ICE. After alignment, they are respectively recorded as 、 、 , input into the fusion model to complete the key feature extraction.

[0036] The fusion model in S100 includes: ,in 、 as well as are the weight coefficients of CT, ICE and DSA respectively. The weight coefficients are generated by the attention mechanism, specifically: , where is the input aligned image data, is the global average pooling, For sigmid( ) function, output weight value 0-1, MLP is a multi-layer perceptron, after obtaining the fused image Then, the kidney tumor area is segmented to distinguish normal tissue from the lesion area, and the lesion area is used as the key feature, and each point in the lesion area is labeled and recorded as a set Obtain fused images After that, perform basic image segmentation to obtain the number of vascular skeleton branches after segmentation. and the overall volume , according to the mathematical formula Obtain vascular branch density and determine vascular wall displacement based on grayscale values , which is convenient for judging the subsequent ablation path. is the grayscale change in the time dimension, is the spatial gradient, To prevent the zero constant, the vascular branching density and vascular wall displacement are conventional features in addition to the key features.

[0037] S200, completing tumor edge recognition based on the acquired key features and generating an optimal ablation path that avoids key structures;

[0038] In S200, the key structures avoided by the optimal ablation path include blood vessels and tumor areas, and the planning of the optimal ablation path includes the following specific steps:

[0039] The lesion area obtained by S100 is divided into safe area and dangerous area. The doctor determines the starting point of the ablation path. , based on the A-Star path planning model, the cost function of the ablation path is established, specifically: , where From the starting point to the current point The actual cost, is the penalty weight, The risk penalty term is introduced because the distance between the multiple points P on the path and the neighboring point k in the safe zone is too close. Therefore, during ablation, it is easy to touch the tissue or organ in the safe zone. , , where For waypoints The Euclidean distance to the neighboring point k, is the decay function, is the attenuation coefficient, is the set of all points in the labeled lesion area, indicating whether the point belongs to the safe area or the dangerous area. For Adjacent points, in the cost function After confirmation, initialize the priority queue, starting point Join the queue, the cost = 0, iteratively expands adjacent points based on the A-Star path planning model, and calculates And update the path, when the end point The purpose of introducing the risk penalty term is to stop when the current point is accessed and to generate a path backtracking. There are dangerous points in the neighborhood of ,but ,Right now Increase , high penalties force the paths to avoid the danger zone during iterations.

[0040] S300, with electromagnetic sensor fusion multi-source image, monitors the deviation of the ablation needle tip and introduces the impedance change rate to dynamically adjust the radiofrequency power;

[0041] In S300, the electromagnetic navigation coordinate system generated by the electromagnetic sensor is E, which is integrated with the fusion image obtained in S100. , specifically: calculate each positioning point collected by the electromagnetic sensor and Corresponding to the cost of each point coordinate, the cost sum is rotated and translated, and the smoothness is constrained to perform regularization terms to complete the offset monitoring of the ablation needle tip. Each positioning point collected by the electromagnetic sensor is the offset point of the tube head end.

[0042] In calculating the relationship between each positioning point collected by the electromagnetic sensor and Corresponding to the cost of each point coordinate, the cost and the cost are rotated and translated, including:

[0043] , where The first Anchor points, For each point coordinate, is the regularization coefficient, is the regularization term, The positioning points collected by the electromagnetic sensor and The cost of the corresponding point coordinates, and in rotation and translation, the positioning point and The closest distance between the corresponding points is recorded as the closest distance from the ablation needle tip to the tumor boundary. , Corresponding points ,After optimization, T includes rotation compensation of 0.5° and translation of (0.2, 0.1, -0, 1) mm to achieve registration and prevent offset.

[0044] Dynamically adjust the RF power, including the following steps:

[0045] according to Calculate the difference between each positioning point collected by the magnetic sensor and The cost and coordinates of the corresponding points are calculated by Obtain the comprehensive error of electromagnetic navigation positioning equipment , combined with the closest distance from the ablation needle tip to the tumor boundary , the introduced impedance change rate , build a control model, specifically: , where is the instrument positioning error, The shortest distance from the ablation needle tip to the tumor boundary. Safety distance threshold 、 、 is the weight coefficient, which is calibrated by clinical data.

[0046] like =1.2mm, =3mm, =5Ω / min, =5mm, then =0.3×sigmoid(-2×1.2)+0.5×tanh(3 / 5)+0.2×5≈-0.1W, so the power of the ablation needle tip is automatically reduced by 0.1W to avoid damaging the blood vessels.

[0047] S400. After surgery, integrate clinical data and pathway points and store them in the clinical database.

[0048] The present invention also provides a renal lesion ablation guidance planning system, comprising:

[0049] Multimodal image fusion module; used to integrate CT, DSA, and ICE image data, complete three-dimensional spatial alignment of images, and extract vascular branch density, vascular wall displacement, and tumor boundaries;

[0050] Path planning module: constructs an ablation path that avoids key structures including blood vessels and tumor areas, and makes real-time path corrections;

[0051] Ablation navigation module; based on the electromagnetic sensor array and impedance detection, it dynamically calculates the device error and regulates the RF power output in combination with the impedance change rate.

[0052] In a specific implementation, this application provides a computer storage medium and a corresponding data processing unit. The computer storage medium is capable of storing a computer program that, when executed by the data processing unit, executes the invention of a method and system for guiding ablation planning for renal lesions provided by the present invention, as well as some or all of the steps in each embodiment. The storage medium may be a magnetic disk, an optical disk, a read-only memory (ROM), or a random access memory (RAM).

[0053] Those skilled in the art will clearly understand that the technical solutions in the embodiments of the present invention can be implemented using computer programs and their corresponding general-purpose hardware platforms. Based on this understanding, the technical solutions in the embodiments of the present invention, or the portion that contributes to the prior art, can be embodied in the form of a computer program, i.e., a software product. This computer program software product can be stored in a storage medium and includes instructions for enabling a device including a data processing unit (such as a personal computer, server, single-chip microcomputer, MCU, or network device) to execute the methods described in various embodiments of the present invention or certain portions of these embodiments.

[0054] The present invention provides a method and system for guiding the planning of ablation of a renal lesion. There are numerous methods and approaches for implementing this technical solution. The foregoing description is merely a preferred embodiment of the present invention. It should be noted that those skilled in the art may make various improvements and modifications without departing from the principles of the present invention, and such improvements and modifications are also within the scope of protection of the present invention. Any components not specified in this embodiment may be implemented using existing technologies.

Claims

1. A method for guiding ablation planning of a renal lesion, characterized in that: The method comprises the following specific steps: S100, based on image acquisition, acquires renal anatomical structure, marks vascular branches and variations, and captures vascular wall micromotion. It also aligns the acquired multi-source images based on a learning algorithm to extract key features. S200, completing tumor edge recognition based on the acquired key features and generating an optimal ablation path that avoids key structures; S300, with electromagnetic sensor fusion multi-source image, monitors the deviation of the ablation needle tip and introduces the impedance change rate to dynamically adjust the radiofrequency power; S400, after surgery, integrates clinical data and pathway points and stores them in the clinical database; In S200, the key structures avoided by the optimal ablation path include blood vessels and tumor areas, and the planning of the optimal ablation path includes the following specific steps: According to the acquired lesion area, the area is divided into safe area and dangerous area, and the starting point of the ablation path is determined , based on the A-Star path planning model, the cost function of the ablation path is established, specifically: , where From the starting point to the current point The actual cost, is the penalty weight, For dangerous penalty items, , where For waypoints The Euclidean distance to the neighboring point k, is the decay function, is the attenuation coefficient, is the set of all points in the labeled lesion area, indicating whether the point k belongs to the safe area or the dangerous area, For adjacent points; In the cost function After confirmation, initialize the priority queue, starting point Join the queue, the cost = 0, iteratively expands adjacent points based on the A-Star path planning model, and calculates And update the path, when the end point Terminate when visited, and backtrack to generate the path.

2. A method for guiding ablation planning of a renal lesion according to claim 1, characterized in that: In the S100, the image acquisition is based on acquiring the anatomical structure of the kidney, marking the vascular branches and variations, and capturing the micro-motion of the vascular wall. , the multi-source images aligned based on the learning algorithm are , the S100 is inputted through the deformation field Finish Every spatial point The displacement vector to The transformation and The transformation relationship between them is, , where is the deformation field, representing each spatial point The displacement vector, is the regularization coefficient, which is used to control the smoothness of the deformation field. The acquisition of renal anatomical structure is based on medical CT, the marking of vascular branches and variations is based on DSA, and the capture of vascular wall micro-motion is based on ICE. After alignment, they are respectively recorded as 、 、 , input into the fusion model to complete the key feature extraction.

3. A method for guiding ablation planning of a renal lesion according to claim 2, characterized in that: The fusion model in S100 includes: ,in 、 as well as are the weight coefficients of CT, ICE and DSA respectively. The weight coefficients are generated by the attention mechanism, specifically: , is the input aligned image data, is the global average pooling, For sigmid( ) function, output weight value 0-1, MLP is a multi-layer perceptron, after obtaining the fused image Then, the kidney tumor area is segmented to distinguish normal tissue from the lesion area, and the lesion area is used as the key feature, and each point in the lesion area is labeled, recorded as .

4. A method for guiding ablation planning of a renal lesion according to claim 1, characterized in that: In the above S300, the electromagnetic navigation coordinate system generated by the electromagnetic sensor is E, and the fusion image obtained in the above S100 is fused , specifically: calculate each positioning point collected by the electromagnetic sensor and Corresponding to the cost of each point coordinate, the cost sum is rotated and translated, and the smoothness is constrained to perform regularization terms to complete the offset monitoring of the ablation needle tip. Each positioning point collected by the electromagnetic sensor is the offset point of the tube head end.

5. A kidney lesion ablation guidance planning method according to claim 4, characterized in that: In calculating the relationship between each positioning point collected by the electromagnetic sensor and Corresponding to the cost of each point coordinate, the cost and the cost are rotated and translated, including: , where The first Anchor points, For each point coordinate, is the regularization coefficient, is the regularization term, The positioning points collected by the electromagnetic sensor and The cost of the corresponding point coordinates, and in rotation and translation, the positioning point and The closest distance between the corresponding points is recorded as the closest distance from the ablation needle tip to the tumor boundary. .

6. A method for guiding ablation planning of a renal lesion according to claim 5, characterized in that: Dynamically adjust the RF power, including the following steps: according to Calculate the difference between each positioning point collected by the magnetic sensor and The cost and coordinates of the corresponding points are calculated by Obtain the comprehensive error of electromagnetic navigation positioning equipment , combined with the closest distance from the ablation needle tip to the tumor boundary , the introduced impedance change rate , build a control model, specifically: , where is the instrument positioning error, The shortest distance from the ablation needle tip to the tumor boundary. is the safety distance threshold, 、 、 is the weight coefficient.

7. A renal lesion ablation guidance planning system, applied to the renal lesion ablation guidance planning method according to any one of claims 1 to 6, characterized in that: The system comprises: Multimodal image fusion module; used to integrate CT and ICE image data, complete three-dimensional spatial alignment of images, and extract vascular branch density, vascular wall displacement, and tumor boundaries; Path planning module: constructs an ablation path that avoids key structures including blood vessels and tumor areas, and makes real-time path corrections; Ablation navigation module; based on the electromagnetic sensor array and impedance detection, it dynamically calculates the device error and regulates the RF power output in combination with the impedance change rate.

Images