CN-121982046-A - Space division radiotherapy sub-target area multi-model automatic sketching method and system

CN121982046ACN 121982046 ACN121982046 ACN 121982046ACN-121982046-A

Abstract

The invention relates to the technical field of medical image processing, in particular to a method and a system for automatically sketching multiple models of a space division radiotherapy sub-target area. The invention is internally provided with a plurality of geometric model libraries such as hexagonal closest packing, face-centered cubic, body-centered cubic, simple cubic and the like, builds a unified space screening processing frame and can flexibly select an optimal space segmentation mode, realizes full-flow automation from medical data loading analysis, target three-dimensional model reconstruction and candidate lattice generation screening to contour calculation optimization and DICOM file synthesis, can greatly reduce labor cost and time cost, and can ensure continuity and consistency of sub-targets in a multi-plane reconstruction view by directly executing geometric calculation and combining high-density sampling to generate smooth closed contours, thereby avoiding the problems of size deviation and space distribution irregularity caused by manual operation and realizing standardization of size consistency and space regularity of the sub-targets.

Inventors

YANG XIONG
Dai Zeyi
SONG HONGBING
ZHANG JUN
RUAN CHANGLI

Assignees

武汉大学人民医院（湖北省人民医院）

Dates

Publication Date: 20260505
Application Date: 20260407

Claims (10)

1. A multi-model automatic sketching method for a space division radiotherapy sub-target area is characterized by comprising the following steps: s1, reading a CT image sequence meeting a preset standard and a radiotherapy data standard structure file containing a sketched macroscopic target area, and analyzing and extracting space coordinate system parameters, pixel information and contour data in the structure file; s2, extracting a two-dimensional contour point set of a source target area on a CT slice, reconstructing a three-dimensional space point cloud model of the source target area, and calculating the geometric centroid of the three-dimensional space point cloud model of the source target area; S3, selecting a target model from a preset geometric model library, generating a candidate target region center dot matrix by taking the geometric centroid as a reference origin, and reserving a center point positioned in the target region through space inclusion screening; s4, calculating intersecting section parameters of the sphere of the sub-target area and the CT slice, converting the center of a section from an original three-dimensional coordinate system of the object to a pixel coordinate system of a CT slice image, and generating a pixel-level contour of the sub-target area; S5, packaging all the generated target area contours into updated radiotherapy data standard structure files according to preset standards; S6, parameter setting, model selection, processing result visualization operation and report checking are performed through a system graphical interface, and result verification and evaluation are performed.
2. The method for automatically sketching a plurality of models of a sub-target area for space division radiotherapy according to claim 1, wherein the parameters of the space coordinate system comprise an image direction cosine, an image position, a pixel interval and an interlayer interval, and the source target area is specifically a source target area designated by a user through a system graphical interface and used for generating the sub-target area.
3. The method for automatically sketching a plurality of space-division radiotherapy sub-target volumes according to claim 1, wherein in step S2, the reconstructing a three-dimensional space point cloud model of a source target volume includes: And extracting and collecting two-dimensional contour points of the source target area on all CT slices, and endowing corresponding Z-axis coordinates to construct a three-dimensional space point cloud model of the source target area, wherein the coordinates of the geometric centroid are specifically the arithmetic average value of all the point coordinates in the three-dimensional space point cloud model.
4. The method according to claim 3, wherein in step S3, the predetermined geometric model library comprises a hexagonal closest packing model, a face-centered cubic model, a body-centered cubic model and a simple cubic model, and the user-settable parameters comprise a lattice constant a and a sphere radius R corresponding to the model, and R < a/2.
5. The method for automatically delineating multiple models of a sub-target region for space-division radiotherapy of claim 4, wherein the method comprises the following steps: the basis vector of the hexagonal closest packing model is defined as an intra-layer basis vector: , interlayer basis vector: wherein The interlayer offset vector is: The lattice point generation formula is that the A layer lattice point is: B layer lattice points: Wherein i, j, k are integer indexes; the basis vector of the face-centered cubic packing model is defined as: , , The centroid offset vector includes: , , the lattice point generation formula is as follows: wherein m=1, 2,3; The basis vector of the body-centered cubic model is defined as: , , body center offset vector: the lattice point generation formula is as follows: ; The basis vector of the simple cubic model is defined as: , , the lattice point generation formula is as follows: 。
6. The method for automatically sketching multiple models of a space-division radiotherapy sub-target zone according to claim 5, wherein in the step S3, the space inclusion screening comprises the steps of performing Delaunay triangulation on the three-dimensional space point cloud to construct a convex hull, removing candidate points outside the convex hull, verifying whether the remaining candidate points are located inside a two-dimensional outline of the target zone through a ray method, and reserving all the candidate points located inside the target zone to form a final sub-target zone center coordinate set.
7. The method of claim 1, wherein in step S4, the generating a pixel-level contour of the sub-target region comprises: s41, calculating the section circle parameters For each sub-target center point Calculating the intersection line of the sphere with each layer of CT section according to the set sphere radius size R of the target region, and setting the Z-direction coordinate as Z If (if) Radius of cross-section circle ; S42, coordinate transformation Converting the circle center of the cross section from the original three-dimensional coordinate system of the object to a CT slice pixel coordinate system to obtain the circle center under the pixel coordinate system The transformation formula is expressed as solving a system of linear equations: Wherein the method comprises the steps of And Is the cosine of the direction of the image, For the offset in the original three-dimensional coordinate system of the object, Offset for pixel coordinates; S43, contour generation Based on the circle center under the pixel coordinate system Preset pixel radius r Generating a discrete point set by adopting a preset sampling interval, and fitting the discrete point set into a circular polygon contour of a sub-target area by combining a spline interpolation algorithm; s44, consistency optimization And sequencing the contours of the same sub-target region on adjacent CT slices, and correcting the contour shape deviation by adopting a moving average algorithm to ensure the continuity and consistency of the contours of the sub-target region.
8. The method of claim 1, wherein in step S5, the packaging is an updated standard structure file of radiotherapy data, and the method comprises: inheriting the object, examination and sequence identifier of the original CT image, establishing a complete reference sequence chain, and adding specific SOP instance references to the source CT image for each contour block.
9. The method for automatically sketching the multiple models of the sub-target areas of the space-division radiotherapy according to claim 1, wherein in the step S6, the processing result visualization operation comprises the steps of overlapping and browsing two-dimensional slice outlines, rendering three-dimensional target areas and the sub-target area models, synchronously displaying three views of an axial view, a sagittal view and a coronal view, automatically calculating total number, total volume and center point distance distribution of the sub-target areas, and generating a standardized analysis report.
10. An electronic system for implementing a method for automatically delineating multiple models of a sub-target volume of spatially segmented radiotherapy as claimed in any one of claims 1 to 9, comprising: the data interface module is used for importing/exporting DICOM standard files and analyzing data information; The geometric model library and the calculation engine are internally provided with a hexagonal closest packing model, a face-centered cubic packing model, a body-centered cubic model and a simple cubic model which are used for executing lattice generation and space screening; the contour generator is used for realizing section calculation, coordinate transformation, contour fitting and consistency optimization; The DICOM synthesizer is used for packaging a standard DICOM-RT Structure file; a graphical user interface that supports parameter settings, model selection, visualization operations, and report viewing.

Description

Space division radiotherapy sub-target area multi-model automatic sketching method and system Technical Field The invention relates to the technical field of medical image processing, in particular to a method and a system for automatically sketching multiple models of a space division radiotherapy sub-target area. Background Space division radiotherapy is a therapeutic technique strategy that can enhance tumor killing while mitigating radiation damage to surrounding normal tissue by creating discrete high dose "peak" regions inside the target region. In current clinical practice, this strategy is largely dependent on the physical operator of radiotherapy manually delineating a large number of spherical or spheroid sub-target volume structures in the treatment planning system. The process has the obvious defects of 1) low efficiency, huge time consumption for manually sketching tens to hundreds of spheres, 2) poor precision and consistency, difficulty in guaranteeing mathematical precision and distribution uniformity of sphere sizes and spatial positions, and 3) lack of theoretical guidance, and difficulty in realizing optimal space filling by manual arrangement. In the prior art, although research attempts are made to dose engrave by using regular geometric sketching as a high-dose target, an integrated automatic generation system which can support various geometric models and realize seamless conversion from medical images to standard treatment plan data is lacking. Disclosure of Invention The technical problem to be solved by the invention is to provide a multi-model automatic drawing method and system for space division radiotherapy sub-target areas, which can realize multi-model adaptation, high efficiency, accuracy and quality-controllable sub-target area drawing. In order to solve the technical problems, the invention adopts the following technical scheme: 1. Space division radiotherapy sub-target area multi-model automatic sketching method The invention provides a multi-model automatic sketching method for a space-division radiotherapy sub-target area, which mainly comprises the following steps: S1, loading and analyzing medical images and structural data, namely reading CT image sequences conforming to the DICOM standard and RT Structure structural files containing the delineated macroscopic target areas, and analyzing and extracting space coordinate system parameters, pixel information and contour data in the structural files; s2, reconstructing a three-dimensional model of the source target region, namely extracting a two-dimensional contour point set of the source target region on a CT slice, reconstructing a three-dimensional space point cloud model of the source target region, and calculating a geometric centroid O of the three-dimensional space point cloud model of the source target region; S3, generating and spatially screening a multi-model candidate target region central lattice, namely selecting a target model from a preset geometric model library, generating a candidate target region central lattice by taking the geometric centroid O as a reference origin, and reserving a central point positioned in the target region through spatial inclusion screening; S4, calculating and optimizing the contour of the high-fidelity sub-target area, namely calculating the intersecting section parameters of the sphere of the sub-target area and the CT slice, converting the center of the section from an original three-dimensional coordinate system of the object to a pixel coordinate system of a CT slice image, and generating a pixel-level contour of the sub-target area; S5, synthesizing a full-association DICOM-RT Structure file, namely packaging all generated target area contours into an updated RT Structure file according to a DICOM standard; S6, performing interactive visualization and space statistical analysis, namely performing parameter setting, model selection, processing result visualization operation and report checking through a system graphical interface, and performing result verification and evaluation. Preferably, the spatial coordinate system parameters comprise an image direction cosine, an image position, a pixel interval and a layer interval, and the source target area is specifically a source target area designated by a user through a system graphical interface and used for generating the sub target area. Preferably, in step S2, the reconstructing the source target three-dimensional space point cloud model includes: And extracting and collecting two-dimensional contour points of the source target area on all CT slices, and endowing corresponding Z-axis coordinates to construct a three-dimensional space point cloud model of the source target area, wherein the coordinates of the geometric centroid are specifically the arithmetic average value of all the point coordinates in the three-dimensional space point cloud model. Preferably, in step S3, the preset geometric model library comprises