Method for high intensity focused ultrasound treatment planning based on three-dimensional lesion mask

Abstract

The application relates to the technical field of image lesion recognition, and discloses a high-intensity focused ultrasound treatment plan generation method based on a three-dimensional lesion mask, which comprises the following steps: obtaining a three-dimensional image and a lesion mask and converting a coordinate system, identifying a treatment area to determine a Y-axis range, generating a treatment layer according to an interlayer spacing and a boundary mode, splitting a processing unit and generating a treatment line and a treatment point, filling a blind area through a point supplementing algorithm after grouping, and finally outputting a structured plan. The method realizes three-dimensional accurate coverage through hierarchical planning, various boundary modes are flexibly adapted to different lesion boundaries, the point supplementing algorithm solves the coverage blind area for special lesions, and the ablation coverage rate and the treatment plan integrity are significantly improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of image lesion recognition, and particularly relates to a high-intensity focused ultrasound treatment plan generation method based on a three-dimensional lesion mask. BACKGROUND

[0002] High-intensity focused ultrasound (HIFU) is a surgical treatment method suitable for solid tumors. Its principle is to focus a high-energy ultrasound beam from the outside of the body to the tumor tissue in the human body, causing high temperature at the focal point, resulting in coagulation necrosis of the tumor tissue, while not affecting the surrounding normal tissue.

[0003] Developing a treatment plan is the first step of HIFU treatment, and it plays a decisive role in the surgical outcome. The traditional treatment plan is developed by the doctor observing the two-dimensional tomographic image of the patient and determining the positional relationship between the lesion and the surrounding organs by imagination, which is highly dependent on the doctor's experience, the process is tedious, the efficiency is low, and there is a risk of repeated treatment or missed areas. Currently, some systems support the generation of treatment plans based on image segmentation results, but they usually lack precise coverage strategies, hierarchical control, boundary avoidance, and focal point arrangement optimization mechanisms, making it difficult to uniformly and accurately cover the lesion volume in three-dimensional space. SUMMARY

[0004] The purpose of the present application is to provide a high-intensity focused ultrasound treatment plan generation method based on a three-dimensional lesion mask. This method achieves precise coverage in three-dimensional space through hierarchical path planning and provides multiple boundary modes to flexibly handle different lesion boundaries, ensuring coverage integrity. At the same time, it introduces a treatment point supplement algorithm to automatically supplement treatment points for special lesions such as narrow and curved ones or coverage blind spots caused by excessive spacing, significantly improving ablation coverage.

[0005] The present application is implemented through the following technical solutions:

[0006] The high-intensity focused ultrasound treatment plan generation method based on a three-dimensional lesion mask comprises the following steps:

[0007] Step 1: Obtain a three-dimensional image and a lesion mask, and perform coordinate system conversion between the image coordinate system, the physical coordinate system, and the treatment coordinate system. Establish a hierarchical treatment plan in a unified physical space coordinate system.

[0008] Step 2: Identify the treatment area, determine the Y-axis coordinate range containing the lesion tissue, and obtain the lowest point coordinate and the highest point coordinate of the lesion in the Y direction.

[0009] Step 3: Calculate the starting point and ending point physical coordinates of the treatment layer according to the preset interlayer distance ΔY and the selected boundary mode, and generate discrete treatment layers.

[0010] Step 4: In the treatment layer unit, the lesion area in the layer is divided into several spatially independent processing units, and each independent processing unit is traversed to generate several parallel X-axis treatment lines along the Z-axis according to the preset boundary mode and line spacing ΔZ;

[0011] Step 5: Based on the preset boundary mode and point spacing ΔX, each treatment line is divided into treatment points, and the treatment points are divided into multiple execution unit groups according to the sequence rule;

[0012] Step 6: Adopting the treatment point supplement algorithm to automatically supplement the lesion area not effectively covered;

[0013] Step 7: All treatment lines are sorted and numbered by Z coordinate from low to high, and the structured treatment plan is output, which is used for visual presentation or called by the backend control module.

[0014] In this scheme, by unifying multi-coordinate system conversion and establishing hierarchical treatment plan, the precise correspondence between treatment plan and clinical operation is realized; by identifying the treatment area and combining the preset interlayer spacing and boundary mode to generate the treatment layer, the spatial range of different lesions can be flexibly adapted, avoiding the incomplete coverage or redundancy problem in traditional manual plan making; the lesion in the treatment layer is divided into independent processing units and parallel X-axis treatment lines are generated along the Z-axis, which can fine path planning for multiple lesions or complex lesion structure, and improve the treatment specificity; based on the point spacing, the treatment points are generated and grouped, which supports the phased triggering of treatment, balances the tissue load and improves the safety and regulation flexibility of treatment rhythm; the treatment point supplement algorithm can automatically fill the coverage blind area that may exist in the conventional generation, significantly improving the integrity of lesion ablation; the structured plan output according to the Z coordinate sorting can be directly called by the backend control module, realizing the data linkage with the treatment equipment, and the overall method gets rid of the excessive dependence on manual experience, greatly improving the efficiency, standardization and accuracy of treatment plan making, which is suitable for various lesion types and image data, provides an efficient and reliable personalized scheme for high-intensity focused ultrasound treatment, and effectively improves the operation safety and success rate.

[0015] Further, since the obtained treatment layer Y coordinate range cannot be divided by ΔY, the first and last two treatment layers cannot cover the lesion area exactly, the boundary mode includes at least one of rounding, extending the last point or two-side difference, and through the combination selection of the three modes, the method can not only cope with the efficient coverage demand of regular shaped lesions, but also handle the boundary problems of special shaped lesions such as narrow and curved, and can be uniformly applied in X, Y and Z dimensions, ensuring the accuracy and consistency of the arrangement of treatment units (layers, lines and points) in three-dimensional space, effectively avoiding the problems of incomplete coverage, treatment redundancy or poor boundary adaptability that may occur in traditional single boundary processing mode.

[0016] Further, when the boundary mode is the rounding mode, the method comprises the following steps:

[0017] To ensure the accurate coverage of the starting boundary of the lesion bottom and avoid the coverage omission of the treatment starting end due to the deviation, the minimum treatment layer Y coordinate is equal to the minimum value of the lesion area in the Y direction, and every ΔY interval selects a treatment layer in the positive direction of the Y axis to ensure the uniformity of the treatment layer spacing, which is helpful to the balanced distribution of focused energy in the lesion area. If the last selected treatment layer is lower than the highest point of the lesion but the distance is less than , the treatment layer is no longer increased; if the distance is greater than , another treatment layer is selected at the interval ΔY. In this way, the treatment redundancy and efficiency loss caused by the additional treatment layer when the top of the lesion exceeds the last layer in a small range are avoided, and the effective coverage of the upper area of the lesion is ensured when the exceeding range is large, preventing the formation of obvious coverage blind area, which is especially suitable for scenarios where the Y direction boundary of the lesion is relatively regular and the uniformity of the treatment rhythm and energy distribution is required.

[0018] Further, when the boundary mode is the extended end point mode, the method comprises the following steps:

[0019] To ensure the accurate anchoring of the starting boundary of the lesion bottom and avoid the coverage omission of the treatment starting end due to the deviation, the minimum treatment layer Y coordinate is equal to the minimum value of the lesion area in the Y direction, and every ΔY interval selects a treatment layer in the positive direction of the Y axis to ensure the uniformity of the treatment layer spacing, which is helpful to the balanced distribution of focused energy in the lesion area. If the last selected treatment layer is lower than the highest point of the lesion but the distance is less than , the treatment layer is no longer increased; if the distance is greater than

[0020] , another treatment layer is selected at the interval ΔY. In this way, the treatment redundancy and efficiency loss caused by the additional treatment layer when the top of the lesion exceeds the last layer in a small range are avoided, and the effective coverage of the upper area of the lesion is ensured when the exceeding range is large, preventing the formation of obvious coverage blind area, which is especially suitable for scenarios where the Y direction boundary of the lesion is relatively regular and the uniformity of the treatment rhythm and energy distribution is required.

[0021] To ensure complete coverage of the lesion from both the top and bottom ends, avoid any area missing due to boundary shrinkage, the minimum treatment layer Y coordinate is not greater than the minimum value of the lesion Y coordinate and the maximum treatment layer Y coordinate is not lower than the maximum value of the lesion Y coordinate, and the distance from the minimum treatment layer to the lesion is equal to the distance from the maximum treatment layer to the lesion, which balances the treatment redundancy at both ends of the lesion through symmetry design, preventing the problem of excessive focusing of energy due to too close distance at one end and weak coverage area due to too far distance at the other end, especially suitable for scenarios where different sensitive tissues adjacent to both ends of the lesion need to be precisely controlled with a safe distance, at the same time, the distance between the first and last treatment layers can be divided by ΔY, and a treatment layer is taken every ΔY between the first and last treatment layers, ensuring uniform arrangement of treatment layers in the overall range, so that the focused energy forms a regular distribution in three-dimensional space.

[0022] Further, in step 4, the processing unit specifically splits the step as follows: by using a connected domain identification algorithm, the lesion in the treatment layer is divided into mutually disconnected regions, any two points in each region can be connected through a path within the region, and there is no spatial intersection with other regions, which not only ensures uniform coverage of treatment points in a single region, but also adjusts the treatment strategy according to the size and shape of different sub-regions, effectively improving the adaptability to complex morphological lesions such as multiple lesions, lobulated or fibrous septa, and avoiding the problem of incomplete coverage or excessive treatment in some areas due to complex lesion structure in the traditional overall processing mode.

[0023] Further, in step 6, when the value is too large or the lesion shape has special structures such as narrow and curved, the regular generated treatment points may not be able to completely cover the entire lesion area. Therefore, the following point supplement algorithm is used to automatically supplement the regions not effectively covered by the treatment points, in order to improve the ablation coverage rate and the integrity of the treatment plan, the specific steps are as follows:

[0024] Step 61: Set a distance threshold t, determine the current generated treatment point coordinate set P, the set of all pixel point coordinates A in the lesion, initialize the uncovered lesion point list M and the supplementary treatment point set N;

[0025] Step 62: For each pixel point a i ∈A, calculate the Euclidean distance between the pixel point a i and all treatment points in the treatment point coordinate set P, which can locate the possible coverage blind area in the regular treatment point arrangement, if all distances are greater than the threshold t, then a i is added to the uncovered lesion point list M, and the corresponding distance is recorded;

[0026] Step 63: Sort the points in the lesion point list M by the distance from the treatment point coordinate set P in priority, take out the farthest point m max , add it to the supplementary treatment point set N, and delete it from the list M; maxpoints whose distance to the set P is less than t;

[0027] Step 64: repeat step 63 until the list M is empty, and obtain the supplementary treatment point set N.

[0028] Further, since the farthest point is the area farthest from the existing treatment point in the lesion, it is the weakest link of coverage, and preferentially supplementing the point can radiate the largest range of uncovered areas with the least new points, significantly improving the coverage efficiency of single point supplement, reducing unnecessary point supplement, avoiding treatment plan redundancy, therefore in step 63, the specific steps are: taking out the point m farthest from the treatment point coordinate set P in the list M max , add it to the supplementary treatment point set N as a new treatment point, calculate the distance of all remaining points in M to the farthest point m max , if the distance is greater than t, the point is kept in the list M, otherwise it is deleted, and finally the treatment point set N that needs to be supplemented is obtained.

[0029] Further, in step 2, the specific steps of identifying the treatment area are: by traversing the three-dimensional image sequence, filtering out the tomographic images containing the lesion tissue, extracting the Y-axis coordinates corresponding to the tomographic images, and determining the minimum and maximum values of the Y coordinate range.

[0030] Further, the three-dimensional image includes at least one of a coronal plane image and a transverse plane image, and the lesion includes at least one of liver cancer and uterine fibroids.

[0031] Compared with the prior art, the present application mainly has the following advantages and beneficial effects:

[0032] 1. The prior art relies heavily on doctors' manual judgment of two-dimensional tomographic images to develop a treatment plan, which is tedious and inefficient, while the present application realizes full-process automation generation based on a three-dimensional lesion mask, reduces manual intervention through unified coordinate system conversion and hierarchical path planning, greatly improves the planning efficiency, and avoids the problem of repeatability caused by differences in artificial experience, ensuring the standardization and consistency of the treatment plan.

[0033] 2. The prior art lacks precise coverage strategies and hierarchical control, making it difficult to uniformly cover the lesion in three-dimensional space and prone to repeated treatment or missed areas, while the present application flexibly adapts to the lesion boundary through three boundary modes, and combines the precise control of interlayer distance ΔY, interline distance ΔZ and point distance ΔX to realize the uniform arrangement of treatment units (layers, lines and points); for special lesions such as narrow and curved ones, the treatment point supplement algorithm automatically identifies and fills the coverage blind area, solving the problem of incomplete coverage in the prior art, while avoiding redundant treatment.

[0034] 3. The prior art has limited processing capacity for multiple lesions or complex lesion structures, while the present application splits the lesions in the treatment layer into spatially independent processing units through regional decomposition processing, supports personalized planning for multiple lesions in combination with the connected domain recognition mechanism, and each unit can be optimized for treatment path, significantly improving the adaptation capacity for complex lesion structures;

[0035] 4. The prior art lacks effective regulation of treatment rhythm, while the present application divides the treatment points into multiple execution unit groups through the treatment point grouping mechanism, supports phased triggering of treatment, balances tissue load, reduces the risk of tissue damage from single treatment, and facilitates doctors to adjust the treatment rhythm according to real-time conditions, improving operation flexibility and safety. BRIEF DESCRIPTION OF DRAWINGS

[0036] The drawings described herein are used to provide further understanding of the embodiments of the present application, constitute a part of the present application, and do not constitute a limitation on the embodiments of the present application. In the drawings:

[0037] Figure 1 is a schematic diagram of a treatment coordinate system;

[0038] Figure 2 is a schematic diagram of the generation of a treatment plan from body→surface→line→point;

[0039] Figure 3 is a flowchart of the supplementary treatment point algorithm;

[0040] Figure 4 is a complete step diagram of the entire treatment plan generation. DETAILED DESCRIPTION

[0041] In order to make the purpose, technical scheme and advantages of the present application more clear and obvious, the present application is further described in detail below in combination with embodiments and drawings, and the illustrative embodiments of the present application and their descriptions are only used to explain the present application, and do not constitute a limitation on the present application.

[0042] EMBODIMENT

[0043] To solve the problems of low efficiency, dependence on artificial experience, incomplete coverage, poor treatment redundancy and repeatability in existing high-intensity focused ultrasound treatment planning, the present embodiment takes the high-intensity focused ultrasound (HIFU) treatment planning for liver cancer lesions as an example, and provides a high-intensity focused ultrasound treatment planning method based on three-dimensional lesion mask, as shown in Figures 1-4 The method comprises the following steps:

[0044] Step 1: Obtain three-dimensional image and lesion mask, and perform conversion of image coordinate system, physical coordinate system and treatment coordinate system, and establish hierarchical treatment plan in a unified physical space coordinate system;

[0045] Specifically, first, a three-dimensional image of a patient's liver (such as a coronal or transverse image) and a lesion mask (marking the three-dimensional spatial range of the lesion area) are acquired, and then the conversion of the image coordinate system, the physical coordinate system and the treatment coordinate system is performed, unified into the physical space coordinate system, to ensure that the coordinates of the treatment plan are consistent with the HIFU device operation coordinates, wherein the HIFU treatment device coordinate system is as shown in the accompanying Figure 1 The human eye observes the three-dimensional gray-scale image from the positive direction to the negative direction of the Y axis.

[0046] In order to uniformly distribute the treatment points in the lesion area and avoid the overlapping volume of the ablation area of each treatment point being too large, and improve the treatment efficiency, the treatment points are selected at the same distance in the X, Y and Z directions when generating the treatment plan, specifically including the following steps.

[0047] Step 2: Identify the treatment area, determine the Y-axis coordinate range containing the lesion tissue, and obtain the lowest point coordinate and the highest point coordinate of the lesion in the Y direction;

[0048] Specifically, as shown in the accompanying Figure 2 , all the tomographic images of the three-dimensional image sequence are traversed, the tomographic images containing the lesion are selected through the lesion mask, the Y-axis coordinates corresponding to these tomographic images are extracted, the Y-coordinate range containing the lesion is determined, and the lowest point coordinate Y fb and the highest point coordinate Y ft of the lesion in the Y direction are obtained.

[0049] Step 3: According to the preset layer spacing ΔY and the selected boundary mode, the starting point and the ending point of the treatment layer are calculated, and the discrete treatment layer is generated;

[0050] Specifically, as shown in the accompanying Figure 2 , based on the Y-axis range of the lesion [Y fb , Y ft ] identified by the treatment area, the preset layer spacing ΔY and the selected boundary mode, the lowest treatment layer coordinate Y tb and the highest treatment layer coordinate Y tt of the lesion are calculated through the following process, and the discrete treatment layer is generated.

[0051] If the Y-direction boundary of the lesion is relatively regular and there is a requirement for uniformity of the treatment rhythm and energy distribution, the selected boundary mode is the rounding mode, and at this time the following equation needs to be met:

[0052] Y tb = Y fb and meets or

[0053] Ymin, Ymax, ΔY, and the boundary mode of the lesion. The lowest treatment layer Y coordinate is equal to the minimum value of the lesion area in the Y direction, and every ΔY interval selects a treatment layer in the positive direction of the Y axis. If the last selected treatment layer is lower than the highest point of the lesion but the distance is less than , then no more treatment layers are added; if the distance is greater than , then another treatment layer is selected at an interval of ΔY;

[0054] If the requirement for the integrity of the top of the lesion is high, the selected boundary mode is the extended end point mode, which requires the following equation to be met:

[0055] Y tb = Y fb , and Y tt > Y ft , and

[0056] The lowest treatment layer Y coordinate is equal to the minimum value of the lesion area in the Y direction, and every ΔY interval selects a treatment layer in the positive direction of the Y axis, so that the highest treatment layer Y coordinate is exactly not lower than the highest point of the lesion;

[0057] If the lesion needs to be precisely controlled at a safe distance from the adjacent different sensitive tissues at both ends, the selected boundary mode is the two-side difference supplement mode, which requires the following equation to be met:

[0058] Y fb -Y tb = Y tt -Y ft

[0059] The lowest treatment layer Y coordinate is not greater than the minimum value of the lesion Y coordinate, and the highest treatment layer Y coordinate is not lower than the maximum value of the lesion Y coordinate. The distance from the lowest treatment layer to the lesion is equal to the distance from the highest treatment layer to the lesion, and the distance between the first and last treatment layers can be divided by ΔY. Every ΔY interval selects a treatment layer between the first and last treatment layers.

[0060] The finally generated discrete treatment layers cover the lesion Y axis range, and the layer spacing is ΔY. The number and distribution of layers are determined by the boundary mode, which ensures that the treatment layers are uniformly arranged in three-dimensional space and adapt to the lesion boundary, providing a hierarchical framework for the generation of treatment lines and treatment points.

[0061] Step 4: As shown in Figure 2 , the lesion area in the treatment layer is divided into several spatially independent processing units. According to the preset boundary mode and line spacing ΔZ, several parallel X-axis treatment lines are generated along the Z-axis at equal distances.

[0062] Specifically, taking a single treatment layer as a processing unit, the three-dimensional lesion mask data corresponding to the treatment layer and the Y-axis coordinate of the treatment layer in the treatment coordinate system are extracted, the intra-layer lesions are divided into several spatially independent processing units (not connected to each other) through connected domain identification, each unit contains the coordinate information of all lesion pixels in the connected domain, for each independent processing unit, the Z-axis coordinates of all lesion pixels are extracted, the minimum and maximum values of the unit in the Z-axis direction are determined, the Z-axis coordinates of the treatment line are calculated according to the preset boundary mode and the line spacing ΔZ, the boundary mode is step 3 (applicable to Z-axis), according to the calculated Z-axis coordinates of the treatment line, a plurality of treatment lines parallel to the X-axis are generated in the current processing unit, each treatment line is a horizontal line extending along the X-axis direction, covering all lesion X-axis ranges in the unit under the Z coordinate.

[0063] For other independent processing units of the current treatment layer, repeat the above process to generate corresponding treatment lines, and finally complete the arrangement of treatment lines for all independent units of the treatment layer.

[0064] Step 5: Based on the preset boundary mode and point spacing ΔX, each treatment line is divided into treatment points, and the treatment points are divided into a plurality of execution unit groups according to the sequence rule;

[0065] Specifically, as shown in Figure 2 , for each generated treatment line (parallel to the X-axis), the physical coordinate range of the lesion region passed through by the treatment line in the X-axis direction is identified through the lesion mask, that is, the starting point (minimum value of X-axis) and the end point (maximum value of X-axis) of the lesion on the treatment line, according to the shape of the lesion in the X-axis direction, select the adaptive mode from the three preset boundary modes, the boundary mode is step 3 (applicable to X-axis), generate treatment point coordinates based on the selected boundary mode and point spacing ΔX, the density and position of the generated treatment points are based on the preset boundary mode strategy and automatic adjustment, to ensure complete coverage within a given range, while avoiding repeated focusing, and finally the generated treatment points are divided into a plurality of execution unit groups according to the sequence rule, supporting phased triggering during treatment, and improving system control flexibility and tissue load balancing.

[0066] Step 6: When the value is too large or the lesion shape has special structures such as narrow and curved, the treatment points generated by the conventional method may not be able to completely cover the entire lesion area. Therefore, the following point supplement algorithm is used to automatically supplement the treatment points for the lesion area not effectively covered by the treatment points, in order to improve the ablation coverage rate and the completeness of the treatment plan;

[0067] Specifically, as shown in Figure 3 , set the distance threshold to t, determine the current generated treatment point coordinate set P and the set A of all pixel points in the lesion, initialize the uncovered lesion point list M and the supplementary treatment point set N. For each lesion pixel point a i ∈A, calculate ai The Euclidean distance between all treatment points in the set P of treatment points is calculated, and if all distances are greater than t, a i is added to M, and the corresponding distance is recorded; the points in list M are sorted by distance, with the farthest point from the current set of treatment points being given priority, which means that the area covered by the current treatment point is the worst, and priority is given to setting a treatment point to maximize coverage of the surrounding area; repeat the following steps until M is empty: take the farthest point m max from M, add it to N as a new treatment point, calculate the distance between all remaining points in M and m max , if the distance is greater than t, it means that the point cannot be covered by the new treatment point m max , continue to remain in M, otherwise it is removed from M; finally, the set of treatment points N that need to be supplemented is obtained.

[0068] Step 7: Sort all treatment lines by Z coordinate from low to high and number, output the structured treatment plan, which is used for visual presentation or called by the backend control module.

[0069] Specifically, all treatment lines of all lesions at all levels are sorted by Z coordinate from low to high and numbered to generate a complete executable treatment plan. In some embodiments, the result can also be packaged in a structured format and can be used for visual presentation or called by the backend control module to realize data linkage with the treatment device.

[0070] In summary, the method realizes accurate coverage of lesions in three-dimensional space through hierarchical planning, multi-boundary mode and automatic point supplementation, avoids dependence on artificial experience, improves planning efficiency and standardization, and is suitable for lesions such as liver cancer and uterine fibroids, providing a reliable personalized solution for HIFU treatment.

[0071] The above specific embodiments further illustrate the purpose, technical solutions and advantages of the present application. It should be understood that the above description is only a specific embodiment of the present application and is not intended to limit the scope of protection of the present application. Any modification, equivalent replacement, improvement, etc. within the spirit and principles of the present application should be included in the scope of protection of the present application.

Claims

1. A method for high intensity focused ultrasound treatment planning based on a three-dimensional lesion mask, characterized in that, The method comprises the following steps: Step 1: obtaining a three-dimensional image and a lesion mask, and performing conversion among an image coordinate system, a physical coordinate system and a treatment coordinate system to establish a hierarchical treatment plan in a unified physical space coordinate system; Step 2: identifying a treatment region, determining a Y-axis coordinate range containing a lesion tissue, and obtaining a lowest point coordinate and a highest point coordinate of the lesion in the Y direction; Step 3: Calculate the start and end physical coordinates of the treatment layer according to the preset layer spacing and the selected boundary mode, and generate the discrete treatment layer. and the selected boundary mode, and generate the discrete treatment layer. Step 4: dividing the lesion region in a treatment layer into a plurality of spatially independent processing units, traversing each independent processing unit, and generating a plurality of treatment lines parallel to the X axis at equal distances along the Z axis according to a preset boundary mode and a line spacing ΔZ; Step 5: dividing each treatment line into treatment points based on a preset boundary mode and a point spacing ΔX, and dividing the treatment points into a plurality of execution unit groups according to a sequence rule; Step 6: automatically supplementing points to a lesion region that is not effectively covered by using a treatment point supplementing algorithm, and the specific steps of the automatic point supplementing are as follows: Step 61: setting a distance threshold t, determining a currently generated treatment point coordinate set P and a lesion pixel point coordinate set A, and initializing an uncovered lesion point list M and a supplement treatment point set N; Step 62: Calculate the Euclidean distance between each pixel point and all the treatment points in the treatment point coordinate set P, if all the distances are greater than the threshold t, the pixel point will be added to the uncovered lesion point list M and the corresponding distance will be recorded. Step 63: Calculate the Euclidean distance between each pixel point and all the treatment points in the treatment point coordinate set P, if all the distances are greater than the threshold t, the pixel point will be added to the uncovered lesion point list M and the corresponding distance will be recorded. Step 64: Calculate the Euclidean distance between each pixel point and all the treatment points in the treatment point coordinate set P, if all the distances are greater than the threshold t Step 63: Sort the points in the lesion point list M according to the priority of being furthest from the set of treatment point coordinates P, and take out the furthest point. Add supplementary healing points to set N, and remove the furthest point from list M. For points whose distance is ≤ t, the specific steps are as follows: Take out the point farthest from the set of treatment point coordinates P in the list M , and add it to the set of supplementary treatment points N as a new treatment point. Calculate the distances between all the remaining points in M and the farthest point . If the distance is greater than t, the point remains in the list M; otherwise, it is deleted from the list. Finally, the set of supplementary treatment points N is obtained. Step 64: repeating step 63 until the list M is empty, and obtaining the supplement treatment point set N; Step 7: sorting and numbering all the treatment lines in ascending order of Z coordinates, and outputting a structured treatment plan, which is used for visual presentation or called by a backend control module.

2. The three-dimensional lesion mask based high intensity focused ultrasound treatment planning method of claim 1, wherein, The boundary mode comprises at least one of rounding, extending an end point or supplementing a difference on both sides.

3. The three-dimensional lesion mask based high intensity focused ultrasound treatment planning method of claim 2, wherein, When the boundary mode is the rounding mode, the following steps are included: The lowest treatment layer Y coordinate is equal to the minimum value of the lesion area in the Y direction, and every interval is selected to a treatment layer, if the last selected treatment layer is lower than the highest point of the lesion but the distance is less than , then no longer increase the treatment layer; if the distance is greater than , then select another treatment layer with an interval of .

4. The three-dimensional lesion mask based high intensity focused ultrasound treatment planning method of claim 2, wherein, When the boundary mode is the extending end point mode, the following steps are included: The lowest treatment layer Y coordinate is equal to the minimum value of the lesion area in the Y direction, and every interval Select a treatment layer, so that the highest treatment layer Y coordinate is just not lower than the highest point of the lesion.

5. The three-dimensional lesion mask based high intensity focused ultrasound treatment planning method of claim 2, wherein, When the boundary mode is the supplementing difference on both sides mode, the following steps are included: The lowest treatment layer Y coordinate is not greater than the minimum value of the lesion Y coordinate and the highest treatment layer Y coordinate is not lower than the maximum value of the lesion Y coordinate, and the distance from the lowest treatment layer to the lesion is equal to the distance from the highest treatment layer to the lesion, and the distance between the first and last treatment layers can be divided evenly, and every treatment layer is taken.

6. The three-dimensional lesion mask based high intensity focused ultrasound treatment planning method of claim 1, wherein, In step 4, the specific splitting step of the processing unit is: dividing the lesion in the treatment layer into regions that are not connected to each other by using a connected domain identification algorithm, and any two points in each region are connected through a path in the region and have no spatial intersection with other regions.

7. The three-dimensional lesion mask based high intensity focused ultrasound treatment planning method of claim 1, wherein, In step 2, the specific steps of identifying the treatment region are: filtering out a tomographic image containing a lesion tissue by traversing a three-dimensional image sequence, and extracting a Y-axis coordinate corresponding to the tomographic image to determine a minimum value and a maximum value of the Y coordinate range.

8. The three-dimensional lesion mask based high intensity focused ultrasound treatment planning method of claim 1, wherein, The three-dimensional image comprises at least one of a coronal plane image and a transverse plane image, and the lesion comprises at least one of liver cancer and uterine fibroids.

Images