A proton or heavy ion radiotherapy dose deviation early warning method, device, equipment, medium and product

Abstract

This application discloses a method, apparatus, device, medium, and product for early warning of proton or heavy ion radiotherapy dose deviation, relating to the field of radiotherapy dose deviation early warning. The method includes: acquiring beam parameters for proton or heavy ion radiotherapy on a patient, as well as planned CT images of the patient's treatment site, planned WET distribution maps, and guiding images for the current treatment session; performing image registration between the planned CT images and the guiding images to obtain a registered image; constructing a beam path on the registered image based on the beam parameters to obtain a superimposed image; obtaining an actual WET distribution map based on the tissue type of each voxel along the beam path in the superimposed image; comparing the actual WET distribution map with the planned WET distribution map to obtain a WET deviation value; and issuing an early warning based on the WET deviation value, thereby achieving early warning of proton or heavy ion radiotherapy dose deviation. This application features physical interpretability, real-time performance, universality, and low cost.

Description

Technical Field

[0001] This application relates to the field of radiotherapy dose deviation early warning, and in particular to a method, device, equipment, medium and product for proton or heavy ion radiotherapy dose deviation early warning. Background Technology

[0002] The general procedure for radiotherapy is as follows: First, simulation and positioning. Medical imaging is performed on the patient under fixed position. Second, radiotherapy planning. The images from the previous step are transferred to a treatment planning system (TPS) software. This software simulates parameters such as the angle, shape, and energy of the radiation rays, and calculates the radiation dose. It simulates the distribution of radiation dose to the tumor and surrounding organs at risk after radiotherapy, using a dose-volume histogram for statistical quantification. The goal is to ensure the tumor receives a sufficient dose while minimizing the radiation dose to surrounding organs. Third, plan execution. After the radiotherapy plan is jointly confirmed by the radiation oncologist and radiation physicist, it is submitted to the radiotherapy equipment for execution. Before execution, the patient is first repositioned on the radiotherapy equipment's treatment bed using the same fixation device as in the "first step simulation positioning." Then, the imaging system carried by the radiotherapy equipment (such as a cone-beam CT imaging system) is activated to quickly image the patient's treatment area without the patient leaving the treatment bed. Next, the rapidly imaged images are registered with the patient's image space used in the "second step radiotherapy planning." Using the positional changes revealed by the registration results, the radiotherapy treatment bed is moved to restore the patient's position to the same position as in the "first step simulation positioning" and the "second step radiotherapy planning." Finally, the patient's radiotherapy plan is invoked, and radiotherapy is administered.

[0003] Traditional radiotherapy uses X-rays to irradiate the tumor site. A characteristic of this type of radiation is that the radiation dose decreases continuously with depth of penetration. By the time it reaches the tumor, the radiation dose has already passed its peak. Proton or heavy ion radiotherapy, however, utilizes the unique Bragg peak characteristic to control the depth and location of the peak dose by adjusting the beam energy intensity, ensuring the dose peak falls within the tumor target area. This results in a reduction in the dose to normal tissues before and after the tumor, thereby improving treatment efficacy while minimizing radiotherapy complications.

[0004] The accuracy of the incident depth of proton or heavy ion beams directly affects the efficacy of radiotherapy. The dose distribution of proton or heavy ion radiotherapy is highly dependent on the range of protons within the patient's body. Proton or heavy ion beams irradiate tumors, creating Bragg peaks of varying depths through a series of beams with different energies. These peaks connect to form a broadened Bragg peak, thus covering the tumor area. The challenge lies in ensuring that each Bragg peak position calculated by the radiotherapy planning system reaches the designated depth during actual irradiation of the patient. The fundamental physical characteristics of the interaction between radioactive particles like protons and heavy ions and matter make proton radiotherapy highly sensitive to changes in tissue composition and propagation path. Therefore, even a tiny change in its propagation path can lead to a shift in the dose deposition location. Furthermore, the high attenuation gradient and highly precise projection characteristics of proton radiotherapy result in disastrous post-treatment outcomes for patients.

[0005] For example, for a 3cm bone (or other object of a certain density and thickness, such as changes in body thickness due to weight changes or accessories on a positioning device) that unexpectedly appears in the incident path (i.e., not present in the software simulation for radiotherapy planning but appearing in the incident path during actual irradiation due to changes in body position, etc.), the change in photon beam dose is only 11%. However, for proton (or heavy ion) beams, the last few Bragg peaks shift forward, resulting in ineffective irradiation of the posterior edge of the tumor, causing off-target effects. This off-target effect means that the tumor irradiation dose does not reach the therapeutic dose, creating a potential risk for future tumor recurrence.

[0006] In current routine clinical practice, proton or heavy ion radiotherapy procedures typically lack the ability to quantitatively assess the deviation between the actual radiation dose and the simulated dose in the original radiotherapy plan online. While current proton or heavy ion radiotherapy procedures involve acquiring online images of the patient before radiotherapy, these images are primarily used for patient positioning correction. Specifically, image registration ensures that the tumor target location reflected in the pre-treatment images coincides with the tumor target location in the radiotherapy plan, thereby guaranteeing correct patient positioning. The process and principle are similar to the image-guided process of traditional X-ray radiotherapy. However, unlike traditional X-ray radiotherapy, proton or heavy ion radiotherapy not only requires image guidance to align the tumor target but also clarifies whether there have been changes in the beam's incident path, such as… Figure 1As shown in section (a), this patient is scheduled for pelvic lymph node irradiation (thin orange line). The radiation plan is to proceed from back to front, passing through the bed board and the skin and muscle layers of the back to reach the tumor area. Image guidance is performed before proton radiotherapy. It can be seen that although the tumor target area (thin orange line) is perfectly matched, the lower left corner (green dashed box) shows an increase in body thickness of 1.75 cm due to the patient's weight gain. If treatment is performed at this point, the dose distribution will be inconsistent with the previous result. Specifically, there will be a shift in the prescribed dose distribution inside and outside the target area. Figure 1 In part (a), the magenta line shows the path range and direction of the proton beam incident (the yellow line is a schematic of the incident direction, with the two field directions being 160° and 200° respectively, passing through the bed plate and the posterior fat layer from below to reach the tumor target area). Figure 1 Part (b) shows the dose distribution during the planning CT (pCT) phase. The orange-red area represents the high-dose region, which can be understood as the dose distribution as perceived by the clinician. To more clearly illustrate the changes in the actual dose distribution after the patient's weight increases, we added a 1.75 cm thick layer of water at the front end of the proton beam. Figure 1 The area within the magenta line in section (c) is used to simulate increased volume thickness. The new dose distribution is obtained after recalculating the proton therapy dose distribution, as shown below. Figure 1 As shown in section (c) of the diagram. The changed actual dose distribution, compared with... Figure 1 The dose distribution shown in section (b) exhibits significant discrepancies. The high-dose zone has shifted, increasing radiation exposure to vital organs such as the rectum and femur. Without correction, this can lead to unintended therapeutic effects and compromised prognosis. Currently, with existing technology, this 1.75cm error (within the green box) receives no warning and can only be observed visually by the radiation oncologist. Failure to carefully observe this error before treatment will result in the tumor region being off-target in the distal direction of the radiation beam.

[0007] Current dose deviation warning methods can be mainly divided into two categories: one is image-guided geometric correction, which focuses on the alignment of organ position and shape, but cannot assess changes in physical range along the incident path; the other is range verification technology based on beam detection and in vivo signals. These methods can provide range information or indirect evidence of in vivo dose deposition to some extent, but usually require specialized detectors, complex post-processing, or have time delays, making it difficult to achieve broad, real-time coverage in every treatment fraction of routine therapy. In summary, the known methods either lack physical interpretability or have limitations in terms of real-time performance, universality, and cost. Therefore, there is an urgent need for a proton or heavy ion radiotherapy dose deviation warning method with physical interpretability, real-time performance, universality, and low cost. Summary of the Invention

[0008] The purpose of this application is to provide a method, device, equipment, medium, and product for early warning of proton or heavy ion radiotherapy dose deviation, which has the characteristics of physical interpretability, real-time performance, universality, and low cost.

[0009] To achieve the above objectives, this application provides the following solution: In a first aspect, this application provides a method for early warning of proton or heavy ion radiotherapy dose deviation, comprising: acquiring beam parameters for proton or heavy ion radiotherapy on a patient, as well as planned CT images of the patient's treatment site, planned WET distribution maps, and guiding images for the current treatment session.

[0010] The planned CT images are registered with the guiding images of the current treatment session to obtain the registered images.

[0011] Based on the beam parameters, a beam path is constructed on the registered image to obtain a superimposed image.

[0012] Based on the tissue type of each voxel along the beam path in the overlay image, the actual WET distribution map is obtained.

[0013] The WET deviation value is obtained by comparing the actual WET distribution map with the planned WET distribution map.

[0014] Early warning is provided based on WET deviation values ​​to achieve early warning of proton or heavy ion radiotherapy dose deviation.

[0015] Secondly, this application provides a proton or heavy ion radiotherapy dose deviation early warning device, including: a data acquisition module, used to acquire the beam parameters for proton or heavy ion radiotherapy to a patient, as well as the planned CT images of the patient's treatment site, the planned WET distribution map, and the guiding images for the current treatment session.

[0016] The image registration module is used to register the planned CT images with the guiding images of the current treatment to obtain the registered images.

[0017] The beam path construction module is used to construct the beam path on the registered image based on the beam parameters to obtain the superimposed image.

[0018] The actual WET distribution map determination module is used to obtain the actual WET distribution map based on the tissue type of each voxel along the beam path in the overlay image.

[0019] The deviation value calculation module is used to compare the actual WET distribution map with the planned WET distribution map to obtain the WET deviation value.

[0020] The early warning module is used to issue early warnings based on WET deviation values, enabling early warning of proton or heavy ion radiotherapy dose deviations.

[0021] Thirdly, this application provides a computer device, including: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the proton or heavy ion radiotherapy dose deviation early warning method described above.

[0022] Fourthly, this application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the aforementioned proton or heavy ion radiotherapy dose deviation early warning method.

[0023] Fifthly, this application provides a computer program product, including a computer program that, when executed by a processor, implements the aforementioned proton or heavy ion radiotherapy dose deviation early warning method.

[0024] According to the specific embodiments provided in this application, this application has the following technical effects: This application provides a method, device, equipment, medium, and product for early warning of proton or heavy ion radiotherapy dose deviation. Range refers to the physical trajectory length of a particle used in radiotherapy that stops in a substance due to energy depletion. Water equivalent thickness (WET) is a standardized measure of range loss, proportionally converting the stopping power of non-uniform tissue to the equivalent thickness of nominal pure water. The logical connection between the two is that when the cumulative WET value along the path equals the nominal water range corresponding to the initial energy of the beam, the particle stops moving and releases the energy peak (Bragg peak). Therefore, WET fluctuations caused by changes in tissue anatomy are the fundamental cause of range deviation in spatial location. Thus, water equivalent thickness can accurately describe the range deviation along the proton or heavy ion radiation path. As a proton beam propagates through matter, it gradually loses energy. The energy loss per unit path is called the stopping power of that matter for the proton beam. The stopping power ratio (SPR) relative to water is usually used to describe the relative effect of matter on proton dose attenuation. To facilitate comparison and accumulation in treatment planning and clinical implementation, the tissue effect at each point along the path is often equivalent to a water layer of corresponding thickness, i.e., the water equivalent thickness. WET (Weighted Equivalent Tolerance) is highly sensitive to the proton range: tiny changes in WET within the beam can cause millimeter-level shifts in the Bragg peak position, directly affecting target dose coverage and the irradiation status of organs at risk. Clinically, factors such as changes in inter-organ spaces, tumor volume, patient positioning, respiratory movements, and changes in fluid or gas content during radiotherapy can all alter the WET distribution along the incident path without changing the irradiation port position, thus producing substantial dosimetric consequences. For procedures where the beam's end point approaches critical organs (especially when using short-range, oblique-incidence beams), even small changes in WET can lead to significant shifts in the Bragg peak, severely impacting dose distribution. Therefore, using WET as a direct monitoring indicator of dose consistency and range variation has clear theoretical advantages: firstly, WET's physical significance based on stopping power directly couples it to proton dose distribution; secondly, WET calculations can utilize pCT, accelerator logs, rapidly reconstructed positional images, or other near-real-time data sources, enabling rapid assessment through SPR accumulation along the incident path. This could potentially provide early warnings of anatomical or path changes that may lead to dose deviations before or during treatment.This application performs image registration between planned CT images and guiding images to obtain a registered image; constructs a beam path on the registered image based on beam parameters to obtain a superimposed image; obtains the actual WET distribution map based on the tissue type of each voxel on the beam path in the superimposed image; compares the actual WET distribution map with the planned WET distribution map to obtain the WET deviation value; and provides early warning based on the WET deviation value to achieve early warning of proton or heavy ion radiotherapy dose deviation. This provides a physically meaningful and operable method for early detection of dose deviation in clinical practice without relying on expensive dedicated detection hardware, featuring real-time performance and low cost; it can directly quantify WET, providing physical interpretability; and it is applicable to proton or heavy ion radiotherapy, possessing universality. Attached Figure Description

[0025] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0026] Figure 1 Image-guided images of a patient before receiving proton therapy.

[0027] Figure 2 This is a flowchart illustrating a method for early warning of proton or heavy ion radiotherapy dose deviation, provided as an embodiment of this application.

[0028] Figure 3 This is a flowchart illustrating a proton or heavy ion radiotherapy dose deviation early warning method provided in another embodiment of this application.

[0029] Figure 4 This is a schematic diagram of the functional modules of a proton or heavy ion radiotherapy dose deviation early warning device provided in an embodiment of this application.

[0030] Figure 5 This is a schematic diagram of the structure of a computer device provided in an embodiment of this application. Detailed Implementation

[0031] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0032] To make the above-mentioned objectives, features and advantages of this application more apparent and understandable, the application will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0033] In one exemplary embodiment, such as Figure 2 As shown, a method for early warning of dose deviation in proton or heavy ion radiotherapy is provided, including: Step 201: Obtaining the beam parameters for proton or heavy ion radiotherapy to the patient, as well as the planned CT images of the patient's treatment site, the planned WET distribution map, and the guiding images for the current treatment session.

[0034] Step 202: Perform image registration between the planned CT image and the guiding image of the current treatment to obtain the registered image.

[0035] Step 203: Construct the beam path on the registered image based on the beam parameters to obtain the superimposed image.

[0036] Step 204: Obtain the actual WET distribution map based on the tissue type of each voxel along the beam path in the superimposed image.

[0037] Step 205: Compare the actual WET distribution map with the planned WET distribution map to obtain the WET deviation value.

[0038] Step 206: Issue an early warning based on the WET deviation value to achieve early warning of proton or heavy ion radiotherapy dose deviation.

[0039] In another exemplary embodiment of this application, the planned CT image is image registered with the medical image to obtain the registered image, specifically including: image preprocessing of the guiding image of the current treatment.

[0040] The preprocessed guiding image is registered with the planned CT image to obtain the registered image.

[0041] In another exemplary embodiment of this application, a beam path is constructed on the registered image based on the beam parameters to obtain a superimposed image. Specifically, this includes determining the incident direction vector of the beam when performing proton or heavy ion radiotherapy on the patient based on the beam parameters.

[0042] A beam reference path is constructed on the registered image based on the incident direction vector and the beam incident point.

[0043] Using the beam reference path as the central axis, a path volume is constructed based on the influence area of ​​the beam in the lateral direction; the lateral direction is the direction perpendicular to the beam reference path.

[0044] The path body is divided into multiple path pieces along the longitudinal direction to obtain a superimposed image; the longitudinal direction is the beam emission direction.

[0045] In another exemplary embodiment of this application, the actual WET distribution map is obtained based on the tissue category of each voxel on the beam path in the superimposed image. Specifically, this includes: segmenting the image within the path body to obtain the tissue category of each voxel in the image within the path body.

[0046] Based on the tissue type of each voxel in the image within the path body, determine the SPR value of each voxel in the image within the path body.

[0047] The actual WET distribution map is obtained based on the SPR values ​​of each voxel in the image within the path body.

[0048] In another exemplary embodiment of this application, the image within the path body is segmented to obtain the tissue category of each voxel in the image within the path body. Specifically, this includes processing the image within the path body using a threshold-based multi-class image segmentation algorithm or a pre-trained deep learning segmentation model to obtain the tissue category of each voxel in the image within the path body. The threshold-based multi-class image segmentation algorithm can be a conventional multi-threshold segmentation method based on CT value (HU) ranges in the art, mapping each voxel within the path body to its corresponding tissue category according to a preset HU threshold range. The tissue categories can be defined with reference to the following standards: air (HU < -900), lung tissue (-900 ≤ HU < -500), adipose tissue (-150 ≤ HU < -50), soft tissue (-50 ≤ HU < 150), dense soft tissue / enhanced tissue (150 ≤ HU < 300), and bone tissue (HU ≥ 300). This method achieves automatic classification of tissues within the path body through the linear mapping relationship between CT value and tissue density (such as the ICRU Report 44 standard), providing a basic input for subsequent WET calculations.

[0049] In another exemplary embodiment of this application, the actual WET distribution map is obtained based on the SPR values ​​of each voxel in the image within the path body. Specifically, the SPR values ​​of each voxel in the image within the path body are integrated relative to the longitudinal length of the path body to obtain the WET value of each voxel.

[0050] The actual WET distribution map is obtained based on the WET values ​​of each voxel.

[0051] This application also provides a more specific embodiment to describe in detail the above-mentioned proton or heavy ion radiotherapy dose deviation early warning method. The implementation of this application relies on an integrated system, including hardware (such as radiotherapy equipment and imaging systems) and software. The overall workflow is divided into the following steps: Step 1: Data preparation and input. Collect planning CT images, planning WET distribution maps, beam parameters, and tumor target contours during the radiotherapy planning phase.

[0052] Step 2: Online Image Acquisition and Preprocessing. Before each treatment, online images (medical images) of the patient's treatment site are acquired using the radiotherapy equipment's in-system imaging to obtain guiding images for that treatment session, and then image preprocessing is performed.

[0053] Step 3: Image registration and path extraction. The online images are rigidly or non-rigidly registered with the planned CT images to align anatomical structures; then, based on the beam path, tissue information along the incident path during treatment is extracted from the registered online image space.

[0054] Step 4: Actual WET Calculation. Based on the CT values ​​from the online images, the WET distribution map of the incident path during treatment is calculated using the HU-stopping power ratio (SPR) conversion model.

[0055] Step 5: WET Comparison and Deviation Assessment. Compare the actual WET distribution map with the planned WET distribution map point by point or by integral, calculate the deviation value, and compare it with the preset tolerance threshold.

[0056] Step 6: Warning Triggering and Feedback. If the deviation value exceeds the preset tolerance threshold, the system will immediately trigger a warning (such as a visual alarm or audible prompt), and may optionally suspend treatment for medical staff to check and adjust.

[0057] Step 7: Iterative Optimization and Recording. The system records WET data and alert events for each treatment for subsequent analysis and plan optimization.

[0058] The entire process can be completed in real time before or during treatment, ensuring timely intervention. The key technical details will be described in detail below, step by step. Figure 3 As shown.

[0059] S1: Data preparation and input.

[0060] Data sources for the plan: Planned CT images, planned WET distribution maps, beam parameters (including beam energy, incident angle, and beam shape), and contour data of the tumor target area and organs at risk are exported from the treatment planning system (TPS). Planned WET distribution map. It is calculated by integrating along the beam path after converting the CT values ​​of the planned CT images to SPR using TPS. The formula is as follows: Where L is the beam path length and SPR(x) is the stopping power ratio at position x.

[0061] Data storage and interface: Planned data is stored in a standard format (such as DICOM) and transmitted to the early warning system in real time via a network interface. The system supports multimodal data fusion to ensure compatibility with online imagery. The early warning system executes steps S2 to S5 below.

[0062] S2: Online image acquisition and preprocessing.

[0063] S2.1: Image Acquisition: Before treatment, acquire two-dimensional or three-dimensional online images of the patient's treatment site using the imaging system built into the radiotherapy equipment (such as cone-beam CT, kV-CT, MRI, or sliding-rail CT). Acquisition can be performed after patient positioning, before treatment begins, or during the intervals between treatments.

[0064] S2.2: Preprocessing steps.

[0065] S2.2.1: Noise and Artifact Correction: Image filtering algorithms (e.g., but not limited to Gaussian filtering, median filtering, or nonlocal mean filtering) are applied to reduce noise; at the same time, model-based correction methods (e.g., but not limited to Monte Carlo simulation, iterative reconstruction algorithms, or deep learning artifact removal networks) can be used to suppress scattering artifacts and cupping artifacts common in cone-beam CT.

[0066] S2.2.2: Image Enhancement: Improve the clarity of tissue boundaries through image processing algorithms such as histogram equalization or contrast stretching, which facilitates subsequent segmentation and registration.

[0067] S2.2.3: Format standardization: Convert online images to the same resolution and coordinate system as planned CT images to ensure calculation consistency.

[0068] S2.3: Image registration and path extraction.

[0069] S2.3.1: Image registration.

[0070] Registration methods: Rigid registration (for postural correction) or non-rigid registration (for organ deformation compensation) can be used. Rigid registration is based on feature point matching (such as mutual information maximization or feature point detection); non-rigid registration uses elastic models (such as B-splines or optical flow) to handle tissue deformation (such as respiratory motion or body weight changes).

[0071] Registration Output: Generates a transformation matrix, aligns the online image with the planned CT image, and outputs the registered online image.

[0072] S2.3.2: Path Extraction. After image registration between online and planned CT images is completed, it is necessary to accurately define the propagation path of the proton or heavy ion beam in the actual anatomical structure and identify the tissue components along the path. This process is a crucial bridge connecting the planned world and the real world.

[0073] Beam path definition: Based on the beam parameters output by the radiotherapy planning system, information such as the beam isocenter, gantry angle, and bed angle are read from the beam parameters to determine the incident direction vector of the beam. Based on this direction vector, an initial geometric straight line path is defined, starting from the skin surface (or fixation device surface) in the registered image, passing through the patient's body in the registered image, and extending to the distal boundary of the target area. This path represents the ideal central axis under conditions of no scattering and energy ambiguity, serving as the beam reference path.

[0074] To more accurately simulate the energy deposition and lateral scattering effects of a real beam, this application physically extends the aforementioned beam reference path by defining a three-dimensional path volume. Specifically: with the beam reference path as the central axis, and based on the field size defined in the beam parameters and the lateral scattering characteristics of the beam, a path volume with a circular, elliptical, rectangular, or custom shape defined according to isodose lines is defined. This path volume represents the main dose-affected area of ​​the beam in the lateral direction; this process is called lateral contour modeling. Next, along the longitudinal depth direction of the beam reference path, the path volume is divided into a series of continuous, thin path slices (voxel segments or path units). Each voxel segment or path unit contains CT value information at the corresponding location in the online image, used for subsequent SPR calculation. This process is called depth direction modeling.

[0075] The defined actual beam path (including the beam reference path and path volume) is overlaid on the registered online images (such as cone-beam CT). Radiation oncologists can visually check through the user interface whether the path correctly passes through the planned tissue structures and avoids organs of danger, achieving a manual secondary verification.

[0076] S2.4: Tissue Segmentation: Within a precise beam path volume (path volume), the different tissue types traversed by the path volume must be automatically and accurately identified and segmented. This is a prerequisite for accurately converting radiographic CT values ​​(HU) into the radiophysical parameter SPR. Inaccurate segmentation leads to incorrect SPR assignment, consequently introducing deviations in the actual WET (Weighted Emissions). Segmentation algorithms (such as thresholding, region growing, or the U-Net deep learning model) are used to automatically segment different tissue density types (such as bone, fat, muscle, etc.) within the path volume. The segmentation results are used to accurately calculate the SPR.

[0077] Specifically, threshold-based multi-class image segmentation algorithms are among the most commonly used and efficient segmentation methods. The system pre-calculates or adaptively calculates a set of CT value thresholds to distinguish different tissues. After threshold-based multi-class image segmentation, for transitional CT value regions or small isolated regions within the path volume, a calculation method such as region growing is used. Starting from seed points on the path volume, the region is expanded based on the connectivity between voxels and the similarity of CT values. This can more accurately segment sub-regions with uniform tissue characteristics and avoid errors caused by volume effects. Simultaneously, combining pre-trained deep learning segmentation models (such as U-Net, V-Net, or their variants) can also be considered for segmentation.

[0078] After tissue segmentation, the system assigns an optimal SPR transformation parameter to each identified tissue type. The SPR transformation parameter can be a more precise tissue-specific HU-SPR look-up table (LUT), or parameters from the process of generating SPR maps using dual-energy CT. This segmentation-assignment process ensures that the conversion from image to physical quantity achieves the millimeter-level accuracy required for clinical applications, laying a solid foundation for subsequent reliable WET comparisons and early warning systems.

[0079] S3: Actual WET calculation.

[0080] S3.1: SPR Model Establishment: The core of the conversion model from CT value to stopping power ratio lies in establishing a robust conversion function that maps the CT value of online images to the corresponding stopping power ratio. SPR is a dimensionless physical quantity that describes the ability of a substance to attenuate proton / heavy ion energy relative to water, and it is the cornerstone of WET calculation.

[0081] This application does not rely on a single conversion model, but instead employs a variety of conversion strategies that have been proven effective in the field to ensure the adaptability and accuracy of the system.

[0082] Empirical calibration curve method: This is the most commonly used and mature method, also known as the lookup table method. It incorporates one or more experimentally determined HU-SPR lookup tables. These lookup tables are generated by scanning standard models (phalanxes) containing various known chemical compositions and densities (such as air, water, polystyrene, and bone equivalent materials), measuring their CT values ​​and actual SPR values, and then fitting these tables together. Based on the HU value of each voxel in the online image, linear or piecewise linear interpolation is performed on this lookup table to obtain its SPR value.

[0083] A derivative method based on dual-energy CT: For advanced radiotherapy centers equipped with dual-energy CT imaging systems, this application utilizes the electron density and effective atomic number information provided by dual-energy CT to calculate SPR more accurately using physical formulas (such as a parametric form of the Bethe-Bloch formula). This method effectively reduces the ambiguity of single-energy CT in distinguishing different tissue components (such as iodine contrast agents and bone), thus providing higher conversion accuracy in principle.

[0084] Planned CT mapping method: When the quality of online images is limited (such as severe noise and artifacts in cone-beam CT), as a supplementary or verification method, the system can map the HU-SPR relationship that has been precisely calibrated in the planned CT to the deformation field obtained by image registration onto the online image space, and assign SPR values ​​to the corresponding voxels.

[0085] S3.2: WET integral along the beam path.

[0086] After obtaining the SPR value of each voxel within the path body, calculate the actual WET distribution map. The process follows the basic principles of radiophysics: integrating the SPR of each voxel within the path volume with respect to the longitudinal length of the path volume (path length). The calculation formula can be expressed as: .in, L actual The longitudinal length of the path body is the actual path length of the beam reference path. SPR actual (x) represents the CT value of the online image.

[0087] To more realistically simulate beam behavior, integration is not simply a linear summation. This application considers the following factors to improve computational fidelity: 1. Path subdivision: The defined beam path (whether central axis or path volume) is divided into sufficiently dense sampling points or thin-layer units to ensure that subtle changes at the tissue interface can be captured.

[0088] 2. Three-dimensional path volume integral: For the defined three-dimensional path volume, the calculation can be further extended. For example, the WET of the central axis can be calculated as the main monitoring indicator, while the WET of multiple parallel sub-paths within the path volume (the parallel sub-paths refer to multiple discrete rays that are off-center from the central axis and parallel to the central axis within the path volume; each parallel sub-path consists of multiple continuous path units distributed along the longitudinal direction) can be calculated to assess the range uncertainty or lateral heterogeneity effects within the beam broadening range.

[0089] The specific steps are as follows: Step A: Construct a 3D path volume or lateral contour model. Determine the beam reference path (central axis). Based on the field parameters defined in the treatment planning system, perform lateral physical expansion of the reference path and execute lateral contour modeling. Define an envelope region on a cross-section perpendicular to the central axis, with the central axis as the origin. A profile The shape and size of this region are determined based on the physical properties of the beam. For example, it can be set to a circle or an ellipse, with a radius... R The lateral broadening σ of the beam Bragg peak at this energy is related (e.g., taking...). R=2σ In certain embodiments, the path volume can also be defined as a rectangle or a custom shape based on isodose lines (such as the 5% isodose line). This path volume spatially represents the main energy deposition and lateral scattering influence area of ​​the pencil beam.

[0090] Step B: Path Volume Discretization and Depth Modeling. Along the longitudinal depth direction of the beam path, the 3D path volume is divided into a series of continuous, sheet-like path units, and depth modeling is performed. The depth sampling step size is set to Δ. z (For example, 1mm-2mm). The first k Path unit V k Defined as in depth z k At this location, the thickness is Δ z The cross-section is A profile The system extracts voxel sets from online images (such as CBCT or CT) based on the registered coordinate transformation relationship, ensuring that each path cell falls within the set. V k The set of CT values ​​for all pixels within the range, specifically: obtaining the path unit. V k The corresponding online image data identifies the area contained within the cross-sectional area of ​​the unit. N Each pixel and its corresponding CT value.

[0091] Step C: Tissue Characteristic Mapping. Since the CT value reflects photon attenuation characteristics, while the proton range depends on the electron density and average excitation energy of the material, an empirical calibration curve method is used to map the tissue characteristics. N The CT values ​​of each pixel are mapped one by one to the SPR of the protons. After the conversion, each path unit is no longer a grayscale image composed of CT values, but is reparameterized into a set of SPRs. S k,1 , S k,2 ， ..., S k,NThis set directly reflects the difference in the blocking ability of each location at this cross-section against the proton beam, providing a physical basis for subsequent calculations of range uncertainty. Among them, N This represents the number of pixels within the cross-section. S k,i Indicates the first k On the path unit, the first i The SPR value of each pixel.

[0092] Step D: Perform three-dimensional path volume integral calculation. Unlike traditional one-dimensional line integrals, this step integrates the statistical characteristics of each path unit along the path.

[0093] Valid path WET ( Calculation: To simulate the influence of the beam's average density on the surrounding tissue at a specific depth, the weighted average of the SPR within that cross-section is calculated and accumulated along the depth: .

[0094] Where M is the total number of path units; is the weighting coefficient based on the Gaussian distribution of the beam (pixels closer to the central axis have higher weights), representing the weighting coefficient of the i-th pixel.

[0095] Step E: Integral of heterogeneity uncertainty ( U het Calculation: To quantify the range uncertainty or lateral heterogeneity caused by lateral scattering, calculate the dispersion (e.g., standard deviation or maximum difference) of the SPR distribution within this cross section and accumulate it along the depth: Alternatively, the integral of standard deviation can be used: .in, Indicates the first k The average blocking ability of a path unit is the arithmetic mean of the SPR values ​​of all pixels in that path unit.

[0096] 3. Real-time Guarantee: Parallel computing (such as GPU acceleration) and optimized algorithms (such as fast integration methods) are employed to ensure that WET calculations are completed within seconds, meeting the real-time requirements of treatment. The calculations are limited to the image area covered by the defined beam path volume, rather than processing the entire 3D image volume, greatly reducing the amount of data computation. The WET integration calculation is inherently highly parallel, utilizing the multi-core architecture of modern processors or graphics processing units for acceleration, achieving calculation speeds in the second or even sub-second range.

[0097] S4: WET comparison and bias assessment.

[0098] The core advantage of this application lies in transforming water equivalent thickness (WET) from a static planning parameter into a dynamic, quantifiable, and comparable monitoring indicator. This step aims to systematically compare planned WET with actual WET and conduct a multi-dimensional, clinically oriented assessment of the deviations between the two, providing a basis for decision-making in accurate early warning.

[0099] Data source synchronization: To ensure the validity of the comparison, first ensure... WET actual and WET plan It is based on the same spatial reference frame and physical foundation. This includes: 1. Path consistency: actual calculation WET actual The beam path used must be geometrically consistent with the beam path generated by the planning system. WET plan The original beam paths are strictly aligned. This is ensured through the aforementioned image registration steps. The WET calculation here does not simply extract the central axis data, but rather performs Gaussian weighted integration on all parallel sub-paths within the path body to obtain the effective WET.

[0100] 2. Align the starting and ending points of integration: Clearly define the starting point (e.g., the beam incident surface) and ending point (e.g., the distal boundary of the clinical target volume CTV) of WET integration to ensure that they correspond completely in an anatomical sense and avoid systematic errors introduced by different integration ranges.

[0101] Under the premise of ensuring that the actual calculation path is strictly aligned with the planned beam path and that the integration start and end points (such as from the body surface to the far end of the target area) are completely consistent, this application adopts the following three parallel WET comparison and graded warning logics. Specifically, the comparison is performed on the following three spatial dimensions, and multiple quantitative indicators are flexibly used for numerical representation of each dimension: 1. "Macro-range warning" based on global path comparison.

[0102] The macroscopic absolute deviation of the entire path body is calculated using the following two formulas. ) and macro relative deviation ( ).

[0103] .

[0104] .

[0105] It is a most intuitive macroscopic indicator that directly reflects the lengthening or shortening of the equivalent thickness of the water along the overall beam path. This is used to assess the proportion of local anatomical changes in the overall range. The calculated global absolute and relative deviations are input into a multi-level warning threshold system. By comparing the planned WET with the actual WET, the lateral heterogeneity within the beam broadening range is naturally incorporated into the macroscopic indicators, improving the robustness of macroscopic warnings. Tolerance threshold (yellow warning): [e.g., when...] and When the percentage is 3%, the change is considered acceptable, and only a background record is made without interrupting the current workflow or issuing an alert. Action threshold (orange alert): When... or When the incidence rate reaches 3%, a medium-level alert is triggered, prompting medical staff to pay close attention, record the details in the treatment log, and conduct manual assessments during the current or subsequent sessions. Intervention threshold (red alert): When... If this occurs, a high-level alert should be triggered immediately, strongly recommending a halt to the treatment beam and mandating intervention from the radiation oncologist and physicist to reassess the treatment plan or patient positioning. This threshold can also be adjusted based on the specific circumstances.

[0106] 2. "Dosimetric effect early warning" based on axial distribution comparison.

[0107] This feature is used to precisely pinpoint the depth at which the deviation occurs and to convert the physical deviation into a clinically significant dose deviation. The system plots the planned and actual WET distributions along the beam depth as two curves and generates an absolute deviation axial distribution curve by subtracting them. Specifically, it refers to the depth of beam penetration. z Plot the "Planned WET Distribution Curve" and the "Actual WET Distribution Curve" with the horizontal axis as the x-axis and the local WET values ​​as the y-axis. Position the two curves at the same depth. z Subtracting the two values ​​at each point generates a continuous absolute deviation curve. The system retrieves curves. The abrupt change in slope can be used to identify the initial depth of changes in anatomical structures. For example, if the curve at... z If the slope suddenly increases at 50mm, it indicates the possible presence of high-density material (such as bone offset entering the path) at that depth. Simultaneously, the depth is calculated. z Standard deviation of lateral deviation within the cross section based on parallel sub-paths When the curve is at a certain depth When a sudden change occurs, the system identifies that there is severe lateral heterogeneity at that depth (such as the beam edge cutting into the bone), thereby accurately locating the deviation depth and quantitatively assessing the risk of lateral scattering at that depth.

[0108] To convert pure physical deviation into dose deviation, the system performs the following conversion calculations: a. The system automatically retrieves radiotherapy plan data and identifies the depth coordinates of the Bragg peak position of the beam in the plan. z peak Subsequently, the system generated from the above Extract from the curve z=z peak The value on the vertical axis corresponding to the coordinate point yields the absolute deviation at Bragg peak. and lateral deviation standard deviation By introducing an expansion factor k (If acceptable) k =1 or 2, representing the confidence interval), combined with the local tissue average relative resistance ability within the depth neighborhood of the planned Bragg peak. Calculate the actual physical range offset: .

[0109] b. Extract the one-dimensional dose gradient along the beam direction from the treatment planning system (TPS). G dose (i.e., the rate of change of dose with depth, %Dose / mm). Then, the local equivalent dose deviation caused by the change in range is calculated: .

[0110] The curves can precisely pinpoint which tissue layers (such as newly formed fat layers or altered lung tissue) contribute significantly to the deviation. Visual comparisons allow for a direct visual identification of the specific depth of the deviation. For example, is it in the superficial fixation segment, the intermediate tissue segment, or the final segment before the target area? Such alerts not only observe the magnitude of the indicator values ​​but also the clinical significance of their positive or negative values. The system is based on... plus or minus sign and Comprehensive clinical early warning based on size: positive deviation ( ): This indicates the firing range is too deep. If the calculated range is... The system indicates that the dose to organs at risk (such as the spinal cord and brainstem) behind the target area exceeds the tolerance limit, triggering a mandatory "Organ Overdose Risk" alarm. Negative bias ( ): This indicates insufficient range. If If the system detects a decrease in the coverage of the distal edge of the target volume (e.g., below 95% of the prescribed dose), it triggers a "target underdose (cold spot) risk" alarm, indicating a potential decrease in tumor control. Healthcare professionals can then decide whether to continue treatment or make minor adjustments / repositioning.

[0111] 3. “Anatomy and Source Tracing Early Warning” based on piecewise integral comparison.

[0112] This dimension aims to assist physicists in quickly pinpointing the anatomical root causes of range deviations. The system automatically divides the entire beam path into several continuous segments based on anatomical structures (e.g., "bedside segment," "surface fat segment," "normal pre-target tissue segment"). Local absolute deviations within each segment are calculated, along with their relative position within the target area. The contribution percentage of a segment is considered. If the contribution percentage of a certain segment exceeds a preset threshold (e.g., 70%), the system will directly output a diagnostic prompt. For example, if the local deviation of the "bed board segment" is not close to 0, a warning will be issued stating "The bed board model or index position is incorrect"; if the contribution of the "body surface segment" is extremely large, a warning will be issued stating "The patient's body thickness or weight has changed significantly".

[0113] S5: Early warning and feedback mechanism.

[0114] Multimodal alarm output: When a red alert is triggered, the specific deviation value is displayed. and ), The graph will display a curve and output specific text diagnostic commands (such as "..."). Exceeding limits is mainly due to thinning of the body's surface fat, posing a risk of excessive amounts endangering organs. A buzzer or voice prompt at a specific frequency will be issued simultaneously to ensure that operators intervene immediately.

[0115] This application uses WET as the core physical quantity, directly establishing its physically interpretable coupling relationship with dose and range. Since this approach is positioned within existing online / offline image enhancement strategies, it can be directly loaded onto existing imaging systems (such as CBCT, sliding-rail CT, MRI, etc.) for real-time WET calculation and early warning, without requiring the addition of expensive hardware detectors. This mechanism can be perfectly inherited and embedded into existing radiotherapy workflows, significantly improving the safety boundaries and efficiency of clinical operations without increasing hardware costs.

[0116] This application uses WET as the core physical quantity, directly coupling it with dose and range to achieve a physically interpretable early warning strategy. Furthermore, this application is designed to support existing online or offline image registration and enhancement strategies. Therefore, this application directly performs real-time WET calculations based on existing imaging technologies (cone-beam CT, sliding-rail CT, MRI, etc.) without requiring additional expensive detectors. This application inherits existing radiotherapy workflows, improving clinical operational efficiency through automatic comparison and early warning.

[0117] The advantages of this application are firstly reflected in its real-time performance. As a real-time augmentation method, this application is computationally fast and can be completed quickly before or during treatment, allowing for timely intervention. The second advantage is accuracy. Based on the SPR model, this application fully considers the physical and biological responses of the beam in the transmission path, accurately mapping path changes. The third advantage is its universality. As an enhancement strategy following the integration of clinical workflows, the design of this application is applicable to various proton and tumor sub-radiotherapy scenarios, including complex anatomical structures. Finally, the advantage lies in its low cost and efficiency. This application directly utilizes existing imaging equipment and implements functionality through algorithms instead of hardware devices, reducing hardware costs.

[0118] This application achieves the following beneficial technical effects: 1. It enables the quantification and early warning of dose deviation. This application transforms the previously unquantifiable and experience-dependent visual observation into a quantitative assessment based on the equivalent thickness of water, with millimeter-level precision. This allows potential deviations in dose deposition to be predicted before or during beam exit, significantly advancing the risk detection point from post-treatment to pre-treatment or during treatment, achieving a fundamental shift from passive acceptance to proactive intervention.

[0119] 2. Improved objectivity and repeatability of clinical decision-making. This application provides the radiotherapy team with a unified and objective basis for decision-making by setting a clear and physically meaningful WET deviation threshold. This effectively avoids inconsistencies in judgment that may arise from differences in experience among different technicians, making the critical step of treatment validation standardized and traceable, and significantly improving the quality control level of the diagnosis and treatment process.

[0120] 3. This application establishes a clinical pathway from geometric imaging to physical dosimetry. It adds a new dosimetric warning function to online imaging, such as cone-beam CT, which is routinely used for body positioning correction. This allows hospitals to access clinically valuable physical dosimetry safety information from existing imaging equipment without additional hardware investment, going beyond geometric position verification. This significantly enhances the technological value and usability of existing equipment.

[0121] 4. A complete closed-loop management process of "monitoring-evaluation-early warning" has been established. This application is not merely a calculation tool; by integrating with existing workflows, it constructs a complete quality management closed loop. It can automatically execute the entire process from image acquisition, WET calculation, deviation comparison to threshold judgment, and ultimately provide clear pass, warning, or stop instructions. This transforms advanced dose warning capabilities into a routine and executable clinical operation, providing a solid technical guarantee for the implementation of precision radiotherapy.

[0122] 5. This application calculates and monitors the WET during the actual treatment process, compares it with the WET and corresponding dose distribution set by the planning system, thereby identifying range changes and dose deviations in real time and triggering early warnings, so as to improve the accuracy and safety of proton or heavy ion radiotherapy.

[0123] 6. This application analyzes online images before radiotherapy to quantitatively assess the WET (Wave Expectation Time) of each proton or heavy ion beam as it passes through the bed board, human tissue, and other areas to reach the tumor target area. The WET is then compared with the WET simulated by the radiotherapy plan, and warnings are issued for cases that exceed the tolerance value.

[0124] 7. In the process of proton or heavy ion radiotherapy for cancer patients, in order to ensure that the radiation dose is accurately deposited at the tumor site according to the path and irradiation depth simulated by the planning system software, this application proposes a proton or heavy ion radiotherapy dose deviation early warning method based on water equivalent thickness to achieve real-time monitoring and early warning of proton or heavy ion radiotherapy dose deviation, based on online images collected for patients before each radiotherapy session and real-time analysis of the difference between the "actual achievable depth of radiation" and the "achievable depth of radiation simulated by the radiotherapy planning system". This method enables real-time monitoring and early warning of proton or heavy ion radiotherapy dose deviation, thereby ensuring the accurate implementation of proton or heavy ion radiotherapy.

[0125] Based on the same inventive concept, this application also provides a proton or heavy ion radiotherapy dose deviation early warning device for implementing the aforementioned proton or heavy ion radiotherapy dose deviation early warning method. The solution provided by this device is similar to the solution described in the above method; therefore, the specific limitations of one or more proton or heavy ion radiotherapy dose deviation early warning device embodiments provided below can be found in the limitations of the proton or heavy ion radiotherapy dose deviation early warning method described above, and will not be repeated here.

[0126] In one exemplary embodiment, such as Figure 4 As shown, a proton or heavy ion radiotherapy dose deviation early warning device is provided, including: a data acquisition module, used to acquire beam parameters for proton or heavy ion radiotherapy to patients, as well as planned CT images, planned WET distribution maps, and medical images of the patient's treatment site.

[0127] The image registration module is used to register planned CT images with medical images to obtain registered images.

[0128] The beam path construction module is used to construct the beam path on the registered image based on the beam parameters to obtain the superimposed image.

[0129] The actual WET distribution map determination module is used to obtain the actual WET distribution map based on the tissue type of each voxel along the beam path in the overlay image.

[0130] The deviation value calculation module is used to compare the actual WET distribution map with the planned WET distribution map to obtain the WET deviation value.

[0131] The early warning module is used to issue early warnings based on WET deviation values, enabling early warning of proton or heavy ion radiotherapy dose deviations.

[0132] In one exemplary embodiment, a computer device is provided, which may be a server or a terminal, and its internal structure diagram may be as follows. Figure 5 As shown, this computer device includes a processor, memory, input / output (I / O) interfaces, and a communication interface. The processor, memory, and I / O interfaces are connected via a system bus, and the communication interface is also connected to the system bus via the I / O interfaces. The processor provides computational and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and a database. The internal memory provides the environment for the operation of the operating system and computer programs in the non-volatile storage media. The database stores proton or heavy ion radiotherapy dose deviation warning data. The I / O interfaces are used for information exchange between the processor and external devices. The communication interface is used for communication with external terminals via a network connection. When the computer program is executed by the processor, it implements a proton or heavy ion radiotherapy dose deviation warning method.

[0133] Those skilled in the art will understand that Figure 5 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0134] In one exemplary embodiment, a computer device is provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the above-described method embodiments.

[0135] In one exemplary embodiment, a computer-readable storage medium is provided storing a computer program that, when executed by a processor, implements the above-described method embodiments.

[0136] In one exemplary embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements the above-described method embodiments.

[0137] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of the relevant data must comply with relevant regulations.

[0138] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments described above. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM).

[0139] The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, etc., and are not limited to these.

[0140] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0141] This document uses specific examples to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the methods and core ideas of this application. Furthermore, those skilled in the art will recognize that, based on the ideas of this application, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. A method for early warning of proton or heavy ion radiotherapy dose deviation, characterized in that, The proton or heavy ion radiotherapy dose deviation early warning method includes: Acquire beam parameters for proton or heavy ion radiotherapy to the patient, as well as planned CT images of the patient's treatment site, planned WET distribution maps, and guiding images for the current treatment session. The planned CT images are registered with the guiding images of the current treatment to obtain the registered images; Based on the beam parameters, a beam path is constructed on the registered image to obtain a superimposed image; Based on the tissue type of each voxel along the beam path in the overlay image, the actual WET distribution map is obtained; The WET deviation value is obtained by comparing the actual WET distribution map with the planned WET distribution map. Early warning is provided based on WET deviation values ​​to achieve early warning of proton or heavy ion radiotherapy dose deviation.

2. The method for early warning of proton or heavy ion radiotherapy dose deviation according to claim 1, characterized in that, The planned CT images are registered with the guiding images of the current treatment session to obtain the registered images, which specifically include: Image preprocessing is performed on the guiding images of the treatment session; The preprocessed guiding image is registered with the planned CT image to obtain the registered image.

3. The method for early warning of proton or heavy ion radiotherapy dose deviation according to claim 1, characterized in that, The beam path includes a beam reference path and a path volume; the beam path is constructed on the registered image according to the beam parameters to obtain a superimposed image, specifically including: The incident direction vector of the beam is determined based on the beam parameters when performing proton or heavy ion radiotherapy on the patient. Construct the beam reference path on the registered image based on the incident direction vector and the beam incident point; With the beam reference path as the central axis, a path volume is constructed based on the influence area of ​​the beam in the lateral direction; the lateral direction is the direction perpendicular to the beam reference path. The path body is divided into multiple path pieces along the longitudinal direction to obtain a superimposed image; the longitudinal direction is the beam emission direction.

4. The proton or heavy ion radiotherapy dose deviation early warning method according to claim 3, characterized in that, Based on the tissue classification of each voxel along the beam path in the overlay image, the actual WET distribution map is obtained, specifically including: The image within the path body is segmented to obtain the tissue category of each voxel in the image within the path body; Based on the tissue type of each voxel in the image within the path body, determine the SPR value of each voxel in the image within the path body; The actual WET distribution map is obtained based on the SPR values ​​of each voxel in the image within the path body.

5. The method for early warning of proton or heavy ion radiotherapy dose deviation according to claim 4, characterized in that, The image within the path body is segmented to obtain the tissue category of each voxel in the image within the path body, specifically including: The images within the path body are processed using a threshold-based multi-class image segmentation algorithm or a pre-trained deep learning segmentation model to obtain the tissue category of each voxel in the image within the path body.

6. The method for early warning of proton or heavy ion radiotherapy dose deviation according to claim 4, characterized in that, The actual WET distribution map is obtained based on the SPR values ​​of each voxel in the image within the path body. Specifically, the SPR values ​​of each voxel in the image within the path body are integrated relative to the longitudinal length of the path body to obtain the WET value of each voxel. The actual WET distribution map is obtained based on the WET values ​​of each voxel.

7. A proton or heavy ion radiotherapy dose deviation early warning device, characterized in that, The proton or heavy ion radiotherapy dose deviation early warning device includes: The data acquisition module is used to acquire the beam parameters for proton or heavy ion radiotherapy to the patient, as well as the planned CT images of the patient's treatment site, the planned WET distribution map, and the guiding images for the current treatment session. The image registration module is used to register the planned CT images with the guiding images of the current treatment to obtain the registered images; The beam path construction module is used to construct the beam path on the registered image based on the beam parameters to obtain the superimposed image. The actual WET distribution map determination module is used to obtain the actual WET distribution map based on the tissue category of each voxel on the beam path in the superimposed image; The deviation value calculation module is used to compare the actual WET distribution map with the planned WET distribution map to obtain the WET deviation value; The early warning module is used to issue early warnings based on WET deviation values, enabling early warning of proton or heavy ion radiotherapy dose deviations.

8. A computer device, comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that the processor executes the computer program to implement the proton or heavy ion radiotherapy dose deviation early warning method according to any one of claims 1-6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by a processor, the computer program implements the proton or heavy ion radiotherapy dose deviation early warning method according to any one of claims 1-6.

10. A computer program product, comprising a computer program, characterized in that, When executed by a processor, the computer program implements the proton or heavy ion radiotherapy dose deviation early warning method according to any one of claims 1-6.

Images