Skull base laser ablation operation path planning method based on rigid ablation device

Through the combination of artificial intelligence big model and surgical approach knowledge base, segmented enclosure box and linear collision detection algorithm, the problem of difficult access to skull base tumor location and identification of important brain tissue is solved, and the rapid skull base laser ablation surgery is achieved Accurate path planning reduces damage to brain tissue and skull.

CN119074213AInactive Publication Date: 2024-12-06ZHEJIANG UNIV OF TECH
View PDF 5 Cites 0 Cited by

Patent Information

Application Number
CN202411190825.4
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-28
Publication Date
2024-12-06
Estimated Expiration
Not applicable · inactive patent

AI Technical Summary

Technical Problem

The location of the skull base tumor is difficult to directly reach, and there are many important brain tissues near the lesions, which makes it difficult to accurately identify and avoid laser ablation surgery. The ablation devices are prone to damage brain tissue, and the approach methods are diverse, making it difficult to damage the skull tissue at least.

Method used

Using the skull base laser ablation surgical path planning method based on rigid ablation devices, identify important skull base tissues through artificial intelligence models, establish an expert knowledge base for surgical approaches, segment key brain tissues and abstract them into segmented enclosure boxes, and use linear collisions. The detection algorithm plans the path to minimize the damage to the skull area.

Benefits of technology

It realizes rapid and accurate skull base laser ablation surgery path planning, effectively avoiding important brain tissues, and provides accurate tools for skull base laser ablation surgery, reducing the damage to brain tissue and skull by ablation devices.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119074213A_ABST
    Figure CN119074213A_ABST
Patent Text Reader

Abstract

A skull base laser ablation operation path planning method based on a rigid ablation device comprises the steps that firstly, modeling is conducted on skull base area key tissues including a cranial nerve CN, a skull base artery blood vessel V, a tumor Tu and a skull S, and the cranial nerve CN and the skull base artery blood vessel V are abstracted into segmented bounding boxes; secondly, the skull is divided into an area S1 capable of being ground and an area S2 incapable of being ground, and a bone window area A of an approach is defined according to priori knowledge of a doctor; and finally, starting from the bone window area A, avoiding the cranial nerve CN, the skull base artery blood vessel V and the non-grindable area S2, minimizing the grindable area S1, solving an approach solution set of the rigid ablation device, and obtaining a recommended path. According to the method, the skull base laser ablation operation path can be quickly and accurately planned, meanwhile, important brain tissue is accurately avoided, and an effective tool is provided for formulating a skull base laser ablation operation scheme.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention belongs to the field of artificial intelligence and medical image processing, and relates to a skull base laser ablation surgery path planning method based on a rigid ablation device. Background Art

[0002] Skull base neurosurgery is difficult and risky. The main reasons are that the complex environment of the skull base makes it difficult to fully expose the lesion during surgery, and that some important brain tissues such as skull base blood vessels and cranial nerves may be damaged during surgery. Laser ablation surgery is an emerging minimally invasive surgical method in recent years. Its principle is to use the different heat resistance of lesions and other tissues to inactivate the lesions at high temperature to achieve the purpose of clearing the lesions. Laser ablation surgery has been successfully used in the field of neurosurgery to treat diseases such as epilepsy and cortical gliomas, and has achieved good results.

[0003] However, laser ablation surgery is still rarely used in skull base neurosurgery, and it faces some major challenges: First, the location of skull base tumors is difficult to reach directly. Skull base tumors such as pituitary tumors and acoustic neuromas are blocked by the skull at the site of onset, making it difficult for ablation devices to reach directly. Secondly, the skull base environment is complex, and there are many important brain tissues near the lesions, such as cranial nerves and skull base blood vessels. Current laser ablation surgery is mainly guided by images, and it is difficult to accurately identify these important brain tissues, and ablation devices can easily damage them. Finally, there are various approaches to skull base neurosurgery. How to choose a suitable approach that can minimize the damage to skull tissue and plan the path for the ablation device to reach the lesion has also become a difficult problem. Summary of the invention

[0004] In order to solve the problems that the location of existing skull base tumors is difficult to reach directly, the presence of various important brain tissues near the lesions makes the ablation device easily damage the brain tissue, and how to minimize the damage to the skull tissue, the ablation device plans the path to the lesion. The present invention provides a skull base laser ablation surgery path planning method based on a rigid ablation device, and constructs a skull base laser ablation surgery path planning model; it can quickly and accurately plan the skull base laser ablation surgery path, while accurately avoiding important brain tissue, and provides an effective tool for the formulation of skull base laser ablation surgery plans.

[0005] The present invention provides the following technical solutions:

[0006] A skull base laser ablation surgery path planning method based on a rigid ablation device comprises the following steps:

[0007] Step 1: Artificial intelligence large model for identifying important skull base tissues

[0008] Taking SAM (Segment Anything Model) as the benchmark, we construct an AI big model by inputting multi-modal and cross-scale images such as T1w, FA, Peaks, CT, and MRA. We use the multi-parameter big model to automatically segment important brain tissues such as skull base blood vessels, skull bones, cranial nerves, brainstem nerves, and tumors, and perform 3D modeling on them.

[0009] Step 2: Establishment of surgical approach expert knowledge base

[0010] The surgical approach to the skull base involves the selection of multiple bone windows. The definition of the bone window comes from some important anatomical landmarks, including sutures, landmarks or skull areas, etc. In particular, the landmarks of the skull can be used to abstract the definition of the bone window. Using CT images as input, the annotated skull landmarks are mapped to individuals by establishing a multi-biological information atlas. At the same time, a surgical approach expert knowledge base is established based on the obtained landmarks. The bone window is abstracted as an area surrounded by multiple landmarks.

[0011] Step 3: Laser ablation device surgical path planning

[0012] First, the obtained key brain tissues are divided into three parts: one is the structure that needs to be avoided, including cranial nerves, skull base blood vessels and non-removable skull areas; the second is the removable skull area; the third is the part that needs to be ablated, which refers to the tumor; the structure that needs to be avoided is abstracted as a segmented bounding box, which can reduce the error caused by inaccurate segmentation of these key tissues; the access bone window is obtained as a set of starting points through step 2, and the part that needs to be ablated, that is, the tumor, is used as the end point. The path set is solved by the straight line collision detection algorithm, and the recommended path is solved by minimizing the removable skull area.

[0013] Furthermore, in step 1, the establishment of an artificial intelligence large model for identifying important skull base tissues includes the following steps:

[0014] 1.1 Overall structure

[0015] Taking SAM (Segment Anything Model) as the benchmark, the overall structure is mainly composed of an encoder and a decoder. The encoder is responsible for extracting multimodal and cross-scale image features and capturing detailed information through a multi-layer convolutional network; the decoder maps these features back to the image space to achieve high-precision segmentation output; in addition, the model also combines an attention mechanism to enhance the focus on important areas and further improve the segmentation effect; the entire architecture is designed to balance efficiency and accuracy and adapt to a variety of image segmentation tasks;

[0016] 1.2 Multimodal Data Prediction

[0017] The prediction output structure of the model includes segmentation mask and category label information. The segmentation mask is obtained through multimodal data input, and the specific area of ​​each object in the image is identified in binary form. The category label indicates the object type. This comprehensive output enables the model to efficiently complete a variety of image segmentation tasks. It is worth noting that the segmentation results of cranial nerve bundles and other nerve bundles are not represented in streamline form, but in the form of areas on the image. This is because the reconstruction of cranial nerves itself is a challenging task, which will bring large cumulative errors.

[0018] Furthermore, in step 2, establishing the surgical entry expert knowledge base includes the following steps:

[0019] 2.1 Automatic identification of key anatomical landmarks at the skull base

[0020] Firstly, based on the individual differences in skull bones due to different biological information, a map consisting of 50 mixed male and female adult CT data, 25 female adult data, and 25 male adult data was constructed;

[0021] When performing landmark mapping, the SyN (Symmetric Normalization) registration algorithm is used. The SyN algorithm is an image registration algorithm based on symmetric transformation. Its core idea is to optimize the transformation fields in two directions at the same time, one from the fixed image to the moving image, and the other from the moving image to the fixed image. This symmetry can improve the accuracy and stability of the registration. Specifically, the SyN algorithm first performs a rough registration at a low resolution, and then gradually increases the resolution. The gradient descent method is used to iteratively optimize the transformation fields in two directions to minimize the difference between the fixed image and the moving image. The key is that the SyN algorithm uses a symmetric diffeomorphic transformation model, which ensures that the transformation is one-to-one and continuously differentiable, thereby ensuring that the topological structure of the transformation will not be destroyed. At the same time, the adaptive step size adjustment strategy also ensures that the algorithm can converge quickly.

[0022] In the multi-atlas mapping, the vector field obtained by the SyN algorithm is and Take the average to get the final vector field

[0023]

[0024] 2.2 Establishment of surgical approach knowledge base combined with anatomical landmarks

[0025] Constructing the surgical entry knowledge base requires defining the bone window defined by the enclosed area composed of anatomical landmarks. For some classic approaches, such as the pterional approach and the subtemporal approach, define the point set P that constitutes these approaches. 1,2,3,...,n .

[0026] Furthermore, in step 3, the laser ablation device surgical path planning includes the following steps:

[0027] 3.1 Definition of important skull base tissues

[0028] The important skull base tissues were divided into three parts and defined respectively: (1) the intangible parts: cranial nerves CN, skull base arteries V and the indelible area S2 of the skull S; (2) the indelible area S1 of the skull S; (3) the ablation target Tu: the tumor;

[0029] 3.2 Starting area selection based on surgical approach knowledge base

[0030] According to the surgical approach expert knowledge base, the position of the ablation target Tu is first located. In order to achieve automation, the bone window area A closest to the position of Tu is selected. This method is also a common method used by simulated neurosurgeons to select surgical approaches.

[0031] 3.3 Rigid Device Path Planning

[0032] First, the cranial nerves CN and the skull base arteries V are abstracted into segmented bounding boxes, which can effectively solve the problem of rigid device collision caused by inaccurate segmentation. Secondly, the skull is divided into an abrasive area S1 and an indelible area S2, and the bone window area A of the approach is defined according to the doctor's prior knowledge. Finally, the calculation starts from the bone window area A, avoids the cranial nerves CN, the skull base arteries V and the indelible area S2, minimizes the abrasive area S1, obtains the approach solution set of the rigid ablation device and obtains the recommended path. The structures that need to be avoided are abstracted into segmented bounding boxes, which can reduce the errors caused by inaccurate segmentation of these key tissues.

[0033] Among them, path planning is performed through straight line collision detection. First, the starting endpoints are defined: Start(x1,y1)∈A and Traget(x1,y1)∈Tu, and their parameter equations are:

[0034] L(t)=A+t(BA),t∈[0,1] (2)

[0035] Each bounding box is composed of the lower left corner C(x min ,y min ) and the upper right corner D(x max ,y max ), where the opposite side is defined as follows:

[0036] Left: (x min ,y min ) to (x min ,y max )

[0037] Right: (x max ,y min ) to (x max ,y max )

[0038] Top: (x min ,y max ) to (x max ,y max )

[0039] Below: (x min ,y min ) to (x max ,y min )

[0040] When judging the intersection, use the vector cross product to judge the intersection. For the parameter points O, P, Q, define:

[0041] cross(O,P,Q)=(P x -O x )(Q y -P y )-(P y -O y )(Q x -O x ) (3)

[0042] So the intersection conditions are as follows:

[0043] 1) The straight line intersects the left side:

[0044] cross(Strat,Target,(x min ,y min ))×cross(Strat,Target,(x min ,y max ))<0

[0045] 2) The straight line intersects the right side:

[0046] cross(Strat,Target,(x max ,y min ))×cross(Strat,Target,(x max ,y max ))<0

[0047] 3) The straight line intersects the upper edge:

[0048] cross((x min ,y max ),(x max ,y max),Start)×cross((x min ,y max ),(x max ,y max ),Target)<0

[0049] 4) The straight line intersects the lower edge:

[0050] cross((x min ,y min ),(x max ,y min ),Start)×cross((x min ,y min ),(x max ,y min ),Target)<0

[0051] Thus, all segmented bounding boxes are traversed to obtain the preliminary path solution set R0;

[0052] Define the final path planning solution set R:

[0053] R = {min(S1)|R0}.

[0054] The beneficial effects of the present invention are: being able to quickly and accurately plan the skull base laser ablation surgical path, while accurately avoiding important brain tissue, and providing an effective tool for the formulation of skull base laser ablation surgical plans. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] Figure 1 It is a structural diagram of the artificial intelligence large model for identifying important skull base tissues.

[0056] Figure 2 It is a flow chart of the overall solution. DETAILED DESCRIPTION

[0057] The present invention will be further described below in conjunction with the accompanying drawings.

[0058] Reference Figure 1 and Figure 2 , a skull base laser ablation surgery path planning method based on a rigid ablation device, comprising the following steps:

[0059] Step 1: Artificial intelligence large model for identifying important skull base tissues

[0060] Taking SAM (Segment Anything Model) as the benchmark, by inputting multi-modal and cross-scale images such as T1w, FA, Peaks, CT, and MRA, an artificial intelligence large model is constructed. The multi-parameter large model can automatically segment important brain tissues such as skull base blood vessels, skull bones, cranial nerves, brainstem nerves, tumors, and perform 3D modeling on them.

[0061] In step 1, the establishment of an artificial intelligence large model for identifying important skull base tissues includes the following steps:

[0062] 1.1 Overall structure

[0063] Taking SAM (Segment Anything Model) as the benchmark, the overall structure is mainly composed of an encoder and a decoder. The encoder is responsible for extracting multimodal and cross-scale image features and capturing detailed information through a multi-layer convolutional network; the decoder maps these features back to the image space to achieve high-precision segmentation output. In addition, the model also combines an attention mechanism to enhance the focus on important areas and further improve the segmentation effect; the entire architecture is designed to balance efficiency and accuracy and adapt to a variety of image segmentation tasks;

[0064] 1.2 Multimodal Data Prediction

[0065] The prediction output structure of the model includes segmentation mask and category label information. The segmentation mask is obtained through multimodal data input, and the specific area of ​​each object in the image is identified in binary form. The category label indicates the object type. This comprehensive output enables the model to efficiently complete a variety of image segmentation tasks. It is worth noting that the segmentation results of cranial nerve bundles and other nerve bundles are not represented in streamline form, but in the form of areas on the image. This is because the reconstruction of cranial nerves itself is a challenging task, which will bring large cumulative errors.

[0066] Step 2: Establishment of surgical approach expert knowledge base

[0067] The surgical approach to the skull base involves the selection of multiple bone windows, and the definition of the bone window comes from some important anatomical landmarks, such as sutures, landmarks, and skull regions. In particular, the landmarks of the skull can be used to abstract the definition of the bone window. Using CT images as input, the annotated skull landmarks are mapped to individuals by establishing a multi-biological information atlas. At the same time, a surgical approach expert knowledge base is established based on the landmarks obtained. The bone window can be abstracted as an area surrounded by multiple landmarks.

[0068] In step 2, establishing the surgical entry expert knowledge base includes the following steps:

[0069] 2.1 Automatic identification of key anatomical landmarks at the skull base

[0070] Firstly, based on the individual differences in skull bones due to different biological information, a map consisting of 50 mixed male and female adult CT data, 25 female adult data, and 25 male adult data was constructed;

[0071] When performing landmark mapping, the SyN (Symmetric Normalization) registration algorithm is used. The SyN algorithm is an image registration algorithm based on symmetric transformation. Its core idea is to simultaneously optimize the transformation fields in two directions, one from fixed image to moving image, and the other from moving image to fixed image. This symmetry can improve the accuracy and stability of registration. Specifically, the SyN algorithm first performs a rough registration at a low resolution, and then gradually increases the resolution. The transformation fields in two directions are iteratively optimized by the gradient descent method to minimize the difference between the fixed image and the moving image. The key is that the SyN algorithm uses a symmetric diffeomorphic transformation model, which ensures that the transformation is one-to-one and continuously differentiable, thereby ensuring that the topological structure of the transformation will not be destroyed. At the same time, the adaptive step size adjustment strategy also ensures that the algorithm can converge quickly.

[0072] In the multi-atlas mapping, the vector field obtained by the SyN algorithm is and Take the average to get the final vector field

[0073]

[0074] 2.2 Establishment of surgical approach knowledge base combined with anatomical landmarks

[0075] Constructing the surgical entry knowledge base requires defining the bone window defined by the enclosed area composed of anatomical landmarks. For some classic approaches, such as the pterional approach and the subtemporal approach, define the point set P that constitutes these approaches. 1,2,3,...,n .

[0076] Step 3: Laser ablation device surgical path planning

[0077] First, the obtained key brain tissue is divided into three parts: one is the structure that needs to be avoided, such as cranial nerves, skull base blood vessels and non-removable skull areas; the second is the removable skull area; the third is the part that needs to be ablated, here specifically refers to the tumor; the structure that needs to be avoided is abstracted as a segmented bounding box, which can reduce the error caused by inaccurate segmentation of these key tissues; the access bone window is obtained as a set of starting points through step 2, and the part that needs to be ablated (tumor) is used as the end point, the path set is solved by the straight line collision detection algorithm, and the recommended path is solved by minimizing the removable skull area.

[0078] In step 3, the laser ablation device surgical path planning includes the following steps:

[0079] 3.1 Definition of important skull base tissues

[0080] The important skull base tissues were divided into three parts and defined respectively: (1) the intangible parts: cranial nerves CN, skull base arteries V and the indelible area S2 of the skull S; (2) the indelible area S1 of the skull S; (3) the ablation target Tu: the tumor;

[0081] 3.2 Starting area selection based on surgical approach knowledge base

[0082] According to the surgical approach expert knowledge base, the position of the ablation target Tu is first located. In order to achieve automation, the bone window area A closest to the position of Tu is selected. This method is also a common method used by simulated neurosurgeons to select surgical approaches;

[0083] 3.3 Rigid Device Path Planning

[0084] First, the cranial nerves CN and skull base arteries V are abstracted as segmented bounding boxes, which can effectively solve the problem of rigid device collision caused by inaccurate segmentation. Secondly, the skull is divided into an abrasive area S1 and an inerasable area S2, and the bone window area A of the approach is defined according to the doctor's prior knowledge. Finally, the calculation starts from the bone window area A, avoids the cranial nerves CN, skull base arteries V and the inerasable area S2, minimizes the removable area S1, obtains the approach solution set of the rigid ablation device and derives the recommended path. The structures that need to be avoided are abstracted as segmented bounding boxes, which can reduce the errors caused by inaccurate segmentation of these key tissues.

[0085] Among them, path planning is performed through straight line collision detection. First, the starting endpoints are defined: Start(x1,y1)∈A and Traget(x1,y1)∈Tu, and their parameter equations are:

[0086] L(t)=A+t(BA),t∈[0,1] (2)

[0087] Each bounding box is composed of the lower left corner C(x min ,y min ) and the upper right corner D(x max ,y max ), where the opposite side is defined as follows:

[0088] Left: (x min ,y min ) to (x min ,y max )

[0089] Right: (x max,y min ) to (x max ,y max )

[0090] Top: (x min ,y max ) to (x max ,y max )

[0091] Below: (x min ,y min ) to (x max ,y min )

[0092] When judging the intersection, we use the vector cross product to judge the intersection. For the parameter points O, P, Q, we define:

[0093] cross(O,P,Q)=(P x -O x )(Q y -P y )-(P y -O y )(Q x -O x ) (3)

[0094] So the intersection conditions can be obtained as follows:

[0095] 1) The straight line intersects the left side:

[0096] cross(Strat,Target,(x min ,y min ))×cross(Strat,Target,(x min ,y max ))<0

[0097] 2) The straight line intersects the right side:

[0098] cross(Strat,Target,(x max ,y min ))×cross(Strat,Target,(x max ,y max ))<0

[0099] 3) The straight line intersects the upper edge:

[0100] cross((x min ,y max ),(x max ,y max ),Start)×cross((x min ,ymax ),(x max ,y max ),Target)<0

[0101] 4) The straight line intersects the lower edge:

[0102] cross((x min ,y min ),(x max ,y min ),Start)×cross((x min ,y min ),(x max ,y min ),Target)<0

[0103] Thus, all segmented bounding boxes are traversed to obtain the preliminary path solution set R0;

[0104] Define the final path planning solution set R:

[0105] R = {min(S1)|R0}.

[0106] The contents described in the embodiments of this specification are merely an enumeration of the implementation forms of the inventive concept and are for illustrative purposes only. The protection scope of the present invention should not be considered to be limited to the specific forms described in this embodiment, and the protection scope of the present invention also extends to equivalent technical means that can be thought of by ordinary technicians in this field based on the inventive concept.

Claims

1. A skull base laser ablation surgical path planning method based on a rigid ablation device, characterized in that: The method comprises the following steps: Step 1: Artificial intelligence large model for identifying important skull base tissues Using SAM as a benchmark, we input T1w, FA, Peaks, CT, and MRA images to build an artificial intelligence model. We use the multi-parameter model to automatically segment the skull base blood vessels, skull bones, cranial nerves, brainstem nerves, and tumor tissues, and perform 3D modeling on them. Step 2: Establishment of surgical approach expert knowledge base Taking CT images as input, the annotated skull landmarks are mapped to individuals by building a multi-biological information atlas. At the same time, a surgical approach expert knowledge base is established based on the obtained landmarks, and the bone window is abstracted as an area surrounded by multiple landmarks. Step 3: Laser ablation device surgical path planning First, the obtained key brain tissue is divided into three parts: one is the structure that needs to be avoided, including cranial nerves, skull base blood vessels and non-removable skull areas; the second is the removable skull area; the third is the part that needs to be ablated, including the tumor; the structure that needs to be avoided is abstracted as a segmented bounding box; the access bone window is obtained through step 2 as a set of starting points, and the part that needs to be ablated is used as the end point, the path set is solved by the straight line collision detection algorithm, and the recommended path is solved by minimizing the removable skull area.

2. A skull base laser ablation surgery path planning method based on a rigid ablation device as claimed in claim 1, characterized in that: In step 1, the establishment of an artificial intelligence large model for identifying important skull base tissues includes the following steps: 1.1 Overall structure Taking SAM as a benchmark, the overall structure consists of an encoder and a decoder. The encoder is responsible for extracting multimodal and cross-scale image features and capturing detailed information through a multi-layer convolutional network; the decoder maps the features back to the image space to achieve high-precision segmentation output; the model also combines an attention mechanism to enhance attention to important areas; 1.2 Multimodal Data Prediction The model's prediction output structure includes segmentation mask and category label information. The segmentation mask is obtained through multimodal data input, and the specific area of ​​each object in the image is identified in binary form. The category label indicates the object type. This comprehensive output enables the model to efficiently complete a variety of image segmentation tasks; the segmentation results of the facial and cranial nerve bundles and other nerve bundles are represented by areas on the image.

3. A skull base laser ablation surgery path planning method based on a rigid ablation device as described in claim 1 or 2, characterized in that: In step 2, establishing the surgical entry expert knowledge base includes the following steps: 2.1 Automatic identification of key anatomical landmarks at the skull base Firstly, based on the individual differences in skull bones due to different biological information, a map consisting of 50 mixed male and female adult CT data, 25 female adult data, and 25 male adult data was constructed; When performing landmark mapping, the SyN registration algorithm is used. In the multi-atlas mapping, the vector field obtained by the SyN algorithm is and Take the average to get the final vector field 2.2 Establishment of surgical approach knowledge base combined with anatomical landmarks Constructing the surgical entry knowledge base requires defining the bone window composed of the enclosed area formed by the anatomical landmarks. For the classic approach, define the approach point set P 1,2,3,...,n .

4. A skull base laser ablation surgery path planning method based on a rigid ablation device as described in claim 1 or 2, characterized in that: In step 3, the laser ablation device surgical path planning includes the following steps 3.1 Definition of important skull base tissues The important skull base tissues were divided into three parts and defined separately: (1) the intangible part: cranial nerves CN, skull base arteries V, and the indelible area S2 of the skull S; (2) the indelible area S1 of the skull S; (3) the ablation target T u : Tumor; 3.2 Starting area selection based on surgical approach knowledge base According to the surgical approach expert knowledge base, the position of the ablation target Tu is first located; in order to achieve automation, the bone window area A closest to the position of Tu is selected; 3.3 Rigid Device Path Planning First, the cranial nerves CN and skull base arteries V are abstracted into segmented bounding boxes. Then, the skull is divided into an abrasive area S1 and an indestructible area S2. The bone window area A of the approach is defined based on the doctor's prior knowledge. Finally, the approach solution set of the rigid ablation device is obtained by avoiding the cranial nerves CN, skull base arteries V and the indestructible area S2, minimizing the abrasive area S1, and obtaining the recommended path. Among them, path planning is performed through straight line collision detection. First, the starting endpoints are defined: Start(x1, y1)∈A and Traget(x1, y1)∈Tu, and their parameter equations are: L(t)=A+t(BA), t∈[0,1](2) Each bounding box is composed of the lower left corner C(x min ,y min ) and the upper right corner D(x max ,y max ), where the opposite side is defined as follows: Left: (x min ,y min ) to (x min ,y max ) Right: (x max ,y min ) to (x max ,y max ) Top: (x min ,y max ) to (x max ,y max ) Below: (x min ,y min ) to (x max ,y min ) When judging the intersection, use the vector cross product to judge the intersection. For the parameter points O, P, Q, define: cross(O,P,Q)=(P x -O x )(Q y -P y )-(P y -O y )(Q x -O x ) (3) So the intersection conditions are as follows: 1) The straight line intersects the left side: cross(Strat,Target,(x min ,y min ))× cross(Strat,Target,(x min ,y max ))<0 2) The straight line intersects the right side: cross(Strat,Target,(x max ,y min ))× cross(Strat,Target,(x max ,y max ))<0 3) The straight line intersects the upper edge: cross((x min ,y max ),(x max ,y max ),Start)× cross((x min ,y max ),(x max ,y max ),Target)<0 4) The straight line intersects the lower edge: cross((x min ,y min ),(x max ,y min ),Start)× cross((x min ,y min ),(x max ,y min ),Target)<0 By traversing all the segmented bounding boxes, we get the preliminary path solution set R0 and define the final path planning solution set R: R = {min(S1)|R0}.

Citation Information

Patent Citations

  • Automatic planning platform for focal cortex dysplasia minimally invasive optical fiber damage path

    CN114748163A

  • Method for identifying skull key points in neurosurgery based on multiple maps

    CN117557808A

  • Quantifying mass effect deformation with structural radiomics in brain tumor patients

    US20180025489A1

  • Automated surgical robot

    US20210128236A1

  • Method and Apparatus for Evaluating Surgical Corridors in the Skull

    US20230419591A1