Dose calculation system, dose quality assurance system and storage medium

Abstract

The embodiment of the present invention discloses a dose calculation system, a dose quality assurance system and a storage medium, wherein the dose calculation system includes a processor, and the processor is configured to process the following steps: obtaining dose-related parameters related to radiotherapy equipment; inputting the dose-related parameters into a trained dose calculation model to obtain output information of the dose calculation model, wherein the dose calculation model is constructed based on a deep learning model; and determining dose distribution information according to the output information. The embodiment of the present invention realizes automatic modeling by constructing and training a dose calculation model based on a deep learning model, does not require the user to manually adjust parameters, has a fast calculation speed and improves the calculation efficiency of dose distribution prediction, and accelerates the modeling process of fast Monte Carlo dose calculation in a radiotherapy system; and the dose calculation model of the present invention can be used for quality assurance of radiation dose, improves calculation efficiency and has high accuracy.

Description

Technical Field

[0001] Embodiments of the present invention relate to the technical field of medical equipment, and in particular to a dose calculation system, a dose quality assurance system, and a storage medium. Background Art

[0002] The Treatment Plan System (TPS) is a medical software product that models the radiation source and the patient and uses an optimization algorithm to determine a set of parameters related to radiotherapy so that the prescribed dose given by the doctor can be achieved when the parameters are executed. With the pursuit of accurate dose calculation, fast Monte Carlo dose calculation in treatment planning systems has become particularly important. The fast Monte Carlo algorithm can calculate at a speed of seconds, which can meet the speed requirements of clinical practice. However, due to the requirement for high precision, the fast Monte Carlo algorithm modeling process is relatively complex, and many parameters need to be adjusted, resulting in a slow determination of the Monte Carlo parameters in the Monte Carlo algorithm, which makes the modeling speed of the Monte Carlo dose algorithm slow. Summary of the Invention

[0003] The embodiments of the present invention provide a computing system, a dose quality assurance system, and a storage medium to accelerate the determination of Monte Carlo parameters in a Monte Carlo algorithm and improve the computing efficiency of dose distribution prediction.

[0004] In a first aspect, an embodiment of the present invention provides a dose calculation system, including a processor, wherein the processor is configured to process the following steps:

[0005] Obtain dose-related parameters related to radiotherapy equipment;

[0006] Inputting dose-related parameters into the trained dose calculation model to obtain output information of the dose calculation model, wherein the dose calculation model is constructed based on a deep learning model;

[0007] Dose distribution information is determined based on the output information.

[0008] In a second aspect, an embodiment of the present invention further provides a dose quality assurance system, including a processor configured to process the following steps:

[0009] Obtaining planning parameters related to the radiotherapy equipment, wherein the planning parameters include machine parameters, radiation field information, and corresponding target dose distribution information required to execute the plan;

[0010] Inputting machine parameters, field information, and at least one target position information related to the radiotherapy equipment into a trained dose calculation model to obtain dose calculation distribution information output by the dose calculation model, wherein the dose calculation model is constructed based on a deep learning model;

[0011] comparing the dose calculation distribution information output by the dose calculation model with the target dose distribution information;

[0012] Determine whether the current dose-related parameters meet the requirements based on the comparison results.

[0013] In a third aspect, an embodiment of the present invention further provides a dose calculation device, comprising:

[0014] Related parameter acquisition module, used to obtain dose-related parameters related to radiotherapy equipment;

[0015] a model output acquisition module, configured to input the dose-related parameters into a trained dose calculation model to obtain output information of the dose calculation model, wherein the dose calculation model is constructed based on a deep learning model;

[0016] The dose distribution determination module is configured to determine dose distribution information according to the output information.

[0017] In a fourth aspect, an embodiment of the present invention further provides a dose quality assurance device, comprising:

[0018] A planning parameter acquisition module is used to obtain planning parameters related to the radiotherapy equipment, wherein the planning parameters include machine parameters, radiation field information, and corresponding target dose distribution information required for executing the plan;

[0019] a dose distribution calculation module, configured to input the machine parameters, field information, and at least one target position information into a trained dose calculation model to obtain dose calculation distribution information output by the dose calculation model, wherein the dose calculation model is constructed based on a deep learning model;

[0020] a dose distribution comparison module, configured to compare the dose calculation distribution information output by the dose calculation model with the target dose distribution information;

[0021] The plan parameter judgment module is used to judge whether the current plan parameters meet the requirements based on the comparison result.

[0022] In a fifth aspect, an embodiment of the present invention further provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the following steps:

[0023] Obtain dose-related parameters related to radiotherapy equipment;

[0024] Inputting the dose-related parameters into a trained dose calculation model to obtain output information of the dose calculation model, wherein the dose calculation model is constructed based on a deep learning model;

[0025] Dose distribution information is determined based on the output information.

[0026] In a sixth aspect, an embodiment of the present invention further provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the following steps:

[0027] Obtaining planning parameters related to the radiotherapy equipment, wherein the planning parameters include machine parameters, radiation field information, and corresponding target dose distribution information required to execute the plan;

[0028] Inputting the machine parameters, field information, and at least one target position information into a trained dose calculation model to obtain dose calculation distribution information output by the dose calculation model, wherein the dose calculation model is constructed based on a deep learning model;

[0029] comparing the dose calculation distribution information output by the dose calculation model with the target dose distribution information;

[0030] Determine whether the current plan parameters meet the requirements based on the comparison result.

[0031] The dose calculation system provided by an embodiment of the present invention obtains dose-related parameters related to radiotherapy equipment; inputs the dose-related parameters into a trained dose calculation model to obtain output information of the dose calculation model, wherein the dose calculation model is constructed based on a deep learning model; determines dose distribution information based on the output information, and realizes automatic modeling by constructing and training a dose calculation model based on the deep learning model, without the need for users to manually adjust parameters, with fast calculation speed and improved computational efficiency of dose distribution prediction. When using the Monte Carlo dose algorithm, the modeling process of fast Monte Carlo dose calculation in the radiotherapy system is accelerated; and the dose calculation model of the present invention can be used for quality assurance of radiation dose, thereby improving computational efficiency and high accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] Figure 1 This is a flowchart of steps executed by a processor in a dose calculation system provided in the first embodiment of the present invention;

[0033] Figure 2 This is a flow chart of steps executed by a processor in a dose calculation system provided in the second embodiment of the present invention;

[0034] Figure 3a is a flowchart of steps executed by a processor in a dose calculation system provided in a third embodiment of the present invention;

[0035] Figure 3b This is a schematic diagram of the structure of a dose calculation model provided by the third embodiment of the present invention;

[0036] Figure 4 This is a flowchart of steps executed by a processor in a dose quality assurance system provided by a fourth embodiment of the present invention;

[0037] Figure 5 This is a schematic structural diagram of a dose calculation device provided by a fifth embodiment of the present invention;

[0038] Figure 6 This is a structural diagram of a dosage quality assurance device provided in Example 6 of the present invention. DETAILED DESCRIPTION

[0039] The present invention will be further described in detail below with reference to the accompanying drawings and examples. It will be understood that the specific embodiments described herein are intended only to illustrate the present invention and are not intended to limit the present invention. It should also be noted that, for ease of description, the accompanying drawings only illustrate portions relevant to the present invention, not all structures.

[0040] Example 1

[0041] Figure 1 This is a flow chart of the steps executed by a processor in a dose calculation system provided by the first embodiment of the present invention. This embodiment is applicable to situations when dose distribution information is calculated. Figure 1 As shown, the steps performed by the processor specifically include:

[0042] S110. Obtain dose-related parameters related to radiotherapy equipment.

[0043] In this embodiment, dose-related parameters can be understood as parameters that affect the dose value. Optionally, the dose-related parameters include machine parameters, field information, and at least one target position information. The field information can be the target area that the radiation points to and passes through, and specifically can be information such as the size and position of the target area. In one embodiment, the field can be configured as a square field of different sizes with the field center on the central axis of the radiation beam. In this case, the field information can refer to the field size. The machine parameters can be machine parameters of the radiotherapy equipment, and can include source parameters, MLC (multi-leaf collimator) parameters, JAW (tungsten gate) parameters, etc. The source parameters can include source position, source size, etc., the MLC parameters can include end face shape, blade position, blade width, leakage coefficient, etc., and the JAW parameters include installation height, etc. The target position information can be understood as the three-dimensional spatial position of the measurement point in the water tank coordinate system. The machine parameters can be determined by the performance of the radiotherapy equipment, and the field information and target position information can be determined according to the radiation requirements.

[0044] S120. Input the dose-related parameters into the trained dose calculation model to obtain output information of the dose calculation model, wherein the dose calculation model is constructed based on a deep learning model.

[0045] This embodiment draws on the framework of deep learning to accelerate the modeling process of the Monte Carlo dose algorithm, uses artificial intelligence prediction methods to adjust the parameters of neurons in the dose calculation model to match the measured water tank curve, realizes automatic modeling of the Monte Carlo algorithm, and performs dose calculation through the automatically constructed dose calculation model, thereby improving the efficiency of determining Monte Carlo parameters and accelerating the model construction process.

[0046] For example, after obtaining the dose-related parameters, the dose-related parameters are input into the trained dose training model to obtain output information of the dose training model. The output information of the dose training model can be data information that can characterize the dose distribution within the radiation field, such as the dose value of each coordinate point in the radiation field.

[0047] S130. Determine dose distribution information according to the output information.

[0048] Optionally, after obtaining the output information of the dose calculation model, the dose distribution information is determined based on the output information of the dose calculation model to achieve a prediction of the dose distribution information. Assuming that the output information is the dose value of each coordinate point in the target radiation field, the dose distribution information can be directly constructed based on the dose value of each coordinate point in the output information. In one embodiment, the dose distribution information is a three-dimensional dose distribution composed of the dose values ​​of at least one target position. Exemplarily, the target position can be the position of any coordinate point within the radiation field.

[0049] The dose calculation model provided by an embodiment of the present invention obtains dose-related parameters; inputs the dose-related parameters into the trained dose calculation model to obtain output information of the dose calculation model, wherein the dose calculation model is constructed based on a deep learning model; and determines dose distribution information according to the output information. By constructing and training the dose calculation model based on the deep learning model, automatic modeling is achieved, and there is no need for the user to manually adjust the parameters. The calculation speed is fast and the computational efficiency of dose distribution prediction is improved. When using the Monte Carlo dose algorithm, the modeling process of fast Monte Carlo dose calculation in the radiotherapy system is accelerated. Moreover, the dose calculation model of the present invention can be used for quality assurance of radiation dose, thereby improving the computational efficiency and accuracy.

[0050] Example 2

[0051] Figure 2 This is a flow chart of the steps executed by the processor in a dose calculation system provided by the second embodiment of the present invention. This embodiment is further optimized based on the above solution. Figure 2 As shown, the method includes:

[0052] S210 , obtaining training sample data, where the training sample data includes different sample dose parameters and sample dose information corresponding to the sample dose parameters.

[0053] In this embodiment, training sample data can be generated based on water tank doses for different field sizes and different machine parameters. Alternatively, a Monte Carlo algorithm can be used to determine sample dose information under multiple sample dose parameters, and training sample data can be determined based on the sample dose information under these multiple sample dose parameters. In this embodiment, the Monte Carlo algorithm can be used to calculate the dose distribution in the water tank, replacing the measured water tank data, to obtain sample dose information at sample dose points. In other words, the sample dose information is the dose information in water determined using the Monte Carlo algorithm. For example, different field sizes and different machine parameters can be unified into different sample dose parameters. The Monte Carlo algorithm in the treatment planning system is pre-calculated to calculate the water tank doses for different field sizes and different machine parameters. The resulting dose points are used as sample dose points, and training sample data is generated based on the dose values ​​at these sample dose points. Using the Monte Carlo algorithm to calculate sample dose information under multiple sample dose parameters avoids the technical issue of data size limitations associated with using actual water tank data. A large number of sample dose points can be obtained, providing sufficient sample training data for the dose calculation model, making the trained dose calculation model more accurate and ensuring that the output of the trained dose calculation model is equivalent to the water tank data. Optionally, all sample dose points can be used as training dose points to obtain training sample data; or some sample dose points can be selected as training dose points to obtain training sample data. Specifically, a combination of each machine parameter, the dose point location and field size of the training dose point, and the calculated water tank dose value can be used as a set of training sample data.

[0054] In one embodiment, the training sample data is determined based on the sample dose information under multiple sample dose parameters, including: selecting some sample dose points corresponding to some sample dose parameters as training dose points, and determining the training sample data based on the training dose points. Preferably, some sample dose points are selected as training dose points, and the training sample data is obtained based on the dose values ​​of the training dose points. Selecting some sample dose points as training dose points can avoid overfitting caused by training all sample dose points as training dose points. In one embodiment, selecting some sample dose points corresponding to some sample dose parameters as training dose points includes: selecting a set proportion of sample dose points as training dose points. Optionally, the set proportion can be any proportion greater than 50%, which can be set according to actual needs, such as selecting 85% of the sample dose points as training dose points.

[0055] S220: Use the training sample data to train the pre-built dose calculation model to obtain a trained dose training model.

[0056] After obtaining the training sample data, the pre-constructed dose calculation model is trained using the training sample data to automatically learn the neuron parameters in the dose calculation model, thereby obtaining a trained dose training model. Optionally, the training sample data is input into the pre-constructed dose calculation model, passing it through multiple layers of neurons, ultimately outputting a predicted dose value for the sample dose point. A loss function is then used to calculate a loss value based on the predicted dose value and the training dose value in the training sample data. Based on the loss value, an optimization algorithm is used to optimize the parameters of each neuron, such as using a gradient descent algorithm to optimize the parameters of each neuron. Ultimately, a dose calculation model is obtained that can accurately calculate the water tank dose under different machine parameters and field sizes.

[0057] On the basis of the above scheme, the steps executed by the processor also include: taking the sample dose points other than the training dose points among the sample dose points corresponding to the sample dose parameters as test dose points, and determining the test sample data based on the test dose points; using the test sample data to verify the trained dose calculation model, and obtaining the trained dose calculation model based on the verification results. Optionally, in order to ensure the accuracy of the dose calculation model, the trained dose calculation model is verified using the test sample data to verify whether the prediction results of the dose calculation model are accurate, and when the verification passes, the trained dose calculation model is used to perform dose prediction. When the verification fails, the neuron parameters in the dose calculation model are adjusted until the dose calculation model passes the verification. Among them, the adjustment of the neuron parameters in the dose calculation model can be: re-acquiring the training sample data, and using the re-acquired training sample data to train the dose calculation model; it can also be by adjusting the loss function in the dose calculation model, etc.

[0058] In one embodiment, the trained dose calculation model is verified using test sample data, and the trained dose calculation model is obtained based on the verification result, including: inputting the test sample data into the dose calculation model to obtain the test prediction result output by the dose model; calculating the difference between the test prediction result and the test sample result corresponding to the test sample data, and when the difference is within a set error range, the trained dose calculation model is obtained. Specifically, the trained dose calculation model is verified using the test sample data to obtain the verification result, which can be: inputting the test dose parameters (such as machine parameters, field information, and at least one target position information) in the test sample data into the trained dose calculation model, obtaining the test prediction result output by the dose calculation model, and comparing the test prediction result with the test sample result (i.e., test dose value) corresponding to the test dose parameters. If the difference between the prediction result and the test dose value is within a set threshold range, the dose calculation model is determined to have passed the verification and the trained dose calculation model is obtained; otherwise, the dose calculation model is determined to have failed the verification. In one embodiment, if the verification results of most of the test sample data meet the requirements, the trained model is used as the dose calculation model. For example, if the verification results of 90%, 92%, or 95% or more of the test sample data meet the requirements, the modeling is completed and the trained model is used as the Monte Carlo dose calculation model.

[0059] S230. Obtain dose-related parameters related to the radiotherapy equipment.

[0060] S240: Input the dose-related parameters into the trained dose calculation model to obtain output information of the dose calculation model.

[0061] S250. Determine dose distribution information according to the output information.

[0062] The embodiment of the present invention obtains training sample data; uses the training sample data to train a pre-built dose calculation model to obtain a trained dose training model, and adjusts the neuron parameters in the dose calculation model through an artificial intelligence prediction method to realize automatic algorithm modeling, thereby improving the efficiency and speed of determining Monte Carlo parameters and accelerating the modeling process of the fast Monte Carlo dose algorithm in the treatment planning system.

[0063] Example 3

[0064] Figure 3a This is a flow chart of a dose prediction method provided by the third embodiment of the present invention. This embodiment provides a preferred embodiment based on the above scheme. Figure 3a As shown, the method includes

[0065] S310 , using a Monte Carlo algorithm to calculate sample dose information under different sample dose parameters, and using dose points associated with the sample dose information as sample dose points.

[0066] Optionally, the sample dose parameters include machine parameters, field information, and at least one target position information. Specific machine parameters, field information, and target position information can be found in the above embodiments and will not be described in detail here.

[0067] S320 , selecting 85% of the sample dose points as training dose points, and the remaining 15% of the sample dose points as testing dose points.

[0068] 85% of the total number of sample dose points may be selected as training dose points, and the remaining sample dose points may be selected as test dose points.

[0069] S330 , generating training sample data based on the training dose points, and using the training sample data to train the dose calculation model to obtain a trained dose calculation model.

[0070] Specifically, the machine parameters are input, the physical coordinates of the training dose point (which can be determined based on the constructed coordinate system) and the field size are obtained, and the training dose point is passed through multiple layers of neurons to ultimately output the dose for that training dose point. The output dose is compared with the dose of the corresponding training dose point in the training sample data and the loss function to calculate the loss value. Based on the loss value, an optimization algorithm is used to optimize the parameters of each neuron, ultimately resulting in a dose calculation model that can accurately calculate the water tank dose under different machine parameters and field sizes.

[0071] Among them, the dose calculation model can be built based on TensorFlow. Figure 3b It is a structural diagram of a dose calculation model provided in the third embodiment of the present invention. Figure 3b The structure of the record prediction model is shown schematically in FIG. Figure 3b As shown in Figure 1, the dose calculation model consists of an input layer, a hidden layer, and an output layer. The input layer data consists of sample dose parameters, such as machine parameters, at least one target position (i.e., the spatial position of the water tank data), and field information (i.e., the field size of the water tank curve). Different training samples require adjustment of the corresponding input layer parameters. The hidden layer consists of multiple layers with varying numbers of neurons. The output layer outputs the dose values ​​of the training dose points in the water tank's spatial coordinate system.

[0072] S340: Generate test sample data based on the test dose points, and use the test sample data to verify the trained dose calculation model.

[0073] Specifically, the difference between the output of the dose calculation model and the test sample results (i.e., dose values) in the test sample data is compared. When the difference is within the allowable error range, the dose calculation model is deemed to be suitable for water tank dose calculation. Optionally, if the verification results of most of the test sample data meet the requirements, the trained model can be used as the dose calculation model. For example, if the verification results of more than 90%, 92%, or 95% of the test sample data meet the requirements, the modeling is completed and the trained model is used as the Monte Carlo dose calculation model.

[0074] S350. Predict the dose distribution based on the verified dose calculation model.

[0075] The validated dose calculation model can be used to predict dose distribution and ensure dose quality, such as applying the dose calculation model to the patient phantom quality management module of the TPS system.

[0076] The embodiment of the present invention uses an artificial intelligence prediction method to adjust the neuron parameters in the dose calculation model to realize automatic algorithm modeling, with high calculation efficiency and fast speed, which accelerates the modeling process of the fast Monte Carlo dose algorithm in the treatment planning system.

[0077] Example 4

[0078] Figure 4 This is a flow chart of the steps executed by the processor in a dose quality assurance system provided by the fourth embodiment of the present invention. This embodiment is applicable to the situation when determining whether the dose-related parameters are ready. Figure 4 As shown, the steps performed by the processor specifically include:

[0079] S410: Obtain planning parameters related to the radiotherapy equipment, where the planning parameters include machine parameters, radiation field information, and corresponding target dose distribution information required to execute the plan.

[0080] In this embodiment, planning parameters are parameters that affect the dose value, specified based on the radiotherapy requirements. These planning parameters include the machine parameters, field information, and corresponding target dose distribution information required to execute the plan. The dose calculation model is applied to the patient phantom quality assurance (QA) module of the treatment planning system to predict the dose distribution. Based on the predicted dose distribution, it is determined whether the planning parameters meet the radiotherapy requirements.

[0081] S420. Input machine parameters, field information, and at least one target position information into the trained dose calculation model to obtain dose calculation distribution information output by the dose calculation model, wherein the dose calculation model is constructed based on a deep learning model.

[0082] After determining the planning parameters, the dose-related parameters are used as input into the trained dose calculation model, resulting in dose calculation distribution information output by the dose calculation model. Dose-related parameters include machine parameters, field information, and at least one target location. The dose calculation distribution information can be a three-dimensional dose distribution composed of dose values ​​at at least one target planning location. For details on the construction and training methods of the dose calculation model, please refer to the above-described embodiments and will not be further elaborated here.

[0083] S430 : Compare the dose calculation distribution information output by the dose calculation model with the target dose distribution information.

[0084] After obtaining the dose calculation distribution information predicted by the dose calculation model, the dose calculation distribution information is compared with the target dose distribution information required for radiotherapy, and the dose difference between the dose calculation distribution information and the target dose distribution information at each target planned position is calculated.

[0085] S440. Determine whether the current dose plan parameters meet the requirements based on the comparison results.

[0086] Whether the requirements are met can be determined by calculating the variance and standard deviation between the dose calculation distribution information and the target dose distribution information or by using the dose volume histogram (DVH). After obtaining the dose difference of the dose values ​​of each target plan position, the dose calculation distribution information obtained based on the plan parameters is judged based on the dose difference of the dose values ​​of each target plan position to determine whether it meets the radiotherapy requirements, thereby determining whether the plan parameters are accurate. Optionally, when the dose difference is less than the set dose threshold, it can be determined that the dose value of the target plan position meets the radiotherapy requirements. When the dose values ​​of most target plan positions (such as 95% of the target plan positions) meet the radiotherapy requirements, it can be considered that the dose plan parameters are appropriate and meet the radiotherapy requirements, and radiotherapy can be performed based on the planned dose plan parameters; otherwise, the plan parameters are adjusted to re-predict the dose distribution until the dose calculation distribution information obtained based on the plan parameters meets the radiotherapy requirements.

[0087] The dose quality assurance system provided by an embodiment of the present invention obtains planning parameters related to radiotherapy equipment; inputs the planning parameters into a trained dose calculation model to obtain dose calculation distribution information output by the dose calculation model, wherein the dose calculation model is constructed based on a deep learning model; compares the dose calculation distribution information output by the dose calculation model with target dose distribution information, and the target dose distribution information corresponds to the dose-related parameters; determines whether the current planning parameters are appropriate based on the comparison results, and realizes automatic modeling by constructing and training a dose calculation model based on the deep learning model, thereby accelerating the modeling process of fast Monte Carlo dose calculation in the radiotherapy system, and using the dose calculation model for quality assurance of radiation dose, thereby improving calculation efficiency and high accuracy.

[0088] Example 5

[0089] Figure 5 This is a schematic diagram of the structure of a dose calculation device provided by the fifth embodiment of the present invention. The dose calculation device can be implemented in software and / or hardware. For example, the dose calculation device can be configured in a dose calculation system. Figure 5 As shown, the apparatus includes a relevant parameter acquisition module 510, a model output acquisition module 520, and a dose distribution determination module 530, wherein:

[0090] Related parameter acquisition module 510, used to obtain dose-related parameters related to radiotherapy equipment;

[0091] A model output acquisition module 520 is used to input dose-related parameters into the trained dose calculation model to obtain output information of the dose calculation model, wherein the dose calculation model is constructed based on the deep learning model;

[0092] The dose distribution determination module 530 is configured to determine dose distribution information according to the output information.

[0093] The embodiment of the present invention obtains dose-related parameters related to the radiotherapy equipment through a related parameter acquisition module; the model output acquisition module inputs the dose-related parameters into the trained dose calculation model to obtain output information of the dose calculation model, wherein the dose calculation model is constructed based on a deep learning model; the dose distribution determination module determines the dose distribution information according to the output information, and realizes automatic modeling by constructing and training the dose calculation model based on the deep learning model, without the need for the user to manually adjust the parameters, with fast calculation speed and improved calculation efficiency of dose distribution prediction. When using the Monte Carlo dose algorithm, the modeling process of fast Monte Carlo dose calculation in the radiotherapy system is accelerated; and the dose calculation model of the present invention can be used for quality assurance of radiation dose, thereby improving calculation efficiency and high accuracy.

[0094] Optionally, based on the above solution, the device further includes a model training module, including:

[0095] A training data acquisition unit is used to acquire training sample data before inputting the dose-related parameters into the trained dose calculation model, wherein the training sample data includes different sample dose parameters and sample dose information corresponding to the sample dose parameters;

[0096] The model training unit is used to train the pre-built dose calculation model using the training sample data to obtain a trained dose calculation model.

[0097] Optionally, based on the above solution, the training data acquisition unit is specifically used to:

[0098] A Monte Carlo algorithm is used to determine sample dose information under multiple sample dose parameters, and training sample data is determined based on the multiple sample dose parameters and the corresponding sample dose information.

[0099] Optionally, based on the above solution, the training data acquisition unit is specifically used to:

[0100] Some sample dose points corresponding to some sample dose parameters are selected as training dose points, and training sample data are determined based on the training dose points.

[0101] Optionally, based on the above solution, the device further includes a model testing module, including:

[0102] a test data acquisition unit, configured to use sample dose points other than training dose points among the sample dose points corresponding to the sample dose parameters as test dose points, and determine test sample data based on the test dose points;

[0103] The model verification unit is used to verify the trained dose calculation model using test sample data, and obtain the trained dose calculation model based on the verification result.

[0104] Optionally, based on the above solution, the model verification unit is specifically used to:

[0105] Input the test sample data into the dose calculation model to obtain the test prediction results output by the dose model;

[0106] The difference between the test prediction result and the test sample result corresponding to the test sample data is calculated. When the difference is within the set error range, the trained dose calculation model is obtained.

[0107] Optionally, based on the above solution, the dose distribution information is a three-dimensional dose distribution composed of dose values ​​of at least one target position.

[0108] Optionally, based on the above solution, the dose-related parameters include machine parameters related to the radiotherapy equipment, field information, and at least one target position information.

[0109] Optionally, based on the above solution, the sample dosage information is dosage information in water determined using a Monte Carlo algorithm.

[0110] The dose calculation device provided in the embodiment of the present invention can execute the steps executed by the processor in the dose calculation system provided in any embodiment of the present invention, and has functional modules and beneficial effects corresponding to the execution steps.

[0111] Example 6

[0112] Figure 6This is a schematic diagram of the structure of a dose quality assurance device provided by the sixth embodiment of the present invention. The dose quality assurance device can be implemented in software and / or hardware. For example, the dose quality assurance device can be configured in a dose quality assurance system. Figure 6 As shown, the apparatus includes a planning parameter acquisition module 610, a dose distribution calculation module 620, a dose distribution comparison module 630, and a planning parameter determination module 640, wherein:

[0113] The planning parameter acquisition module 610 is used to obtain planning parameters related to the radiotherapy equipment. The planning parameters include the machine parameters required to execute the plan, the radiation field information, and the corresponding target dose distribution information.

[0114] A dose distribution calculation module 620 is configured to input machine parameters, field information, and at least one target location information into a trained dose calculation model to obtain dose distribution information output by the dose calculation model, wherein the dose calculation model is constructed based on a deep learning model;

[0115] The dose distribution comparison module 630 is used to compare the dose distribution information output by the dose calculation model with the target dose distribution information;

[0116] The plan parameter judgment module 640 is used to judge whether the current plan parameters meet the requirements based on the comparison results.

[0117] The embodiment of the present invention obtains planning parameters related to radiotherapy equipment through a planning parameter acquisition module; the dose distribution calculation module inputs the planning parameters into a trained dose calculation model to obtain dose calculation distribution information output by the dose calculation model, wherein the dose calculation model is constructed based on a deep learning model; the dose distribution comparison module compares the dose calculation distribution information output by the dose calculation model with target dose distribution information, and the target dose distribution information corresponds to dose-related parameters; the planning parameter judgment module judges whether the current planning parameters meet the requirements based on the comparison results, and realizes automatic modeling by constructing and training a dose calculation model based on a deep learning model, thereby accelerating the modeling process of rapid Monte Carlo dose calculation in the radiotherapy system, and using the dose calculation model for quality assurance of radiation dose, thereby improving calculation efficiency and high accuracy.

[0118] The dose quality assurance device provided in the embodiment of the present invention can execute the steps executed by the processor in the dose quality assurance system provided in any embodiment of the present invention, and has functional modules and beneficial effects corresponding to the execution steps.

[0119] Example 7

[0120] Embodiment 7 of the present invention further provides a computer-readable storage medium having a computer program stored thereon. When the program is executed by a processor, the steps executed by the processor in the dose calculation system provided in the embodiment of the present invention are implemented, including:

[0121] Obtain dose-related parameters related to radiotherapy equipment;

[0122] Inputting dose-related parameters into the trained dose calculation model to obtain output information of the dose calculation model, wherein the dose calculation model is constructed based on a deep learning model;

[0123] determining dose distribution information based on the output information;

[0124] And / or, implementing the steps executed by the processor in the dose quality assurance system provided by the embodiment of the present invention includes:

[0125] Obtaining planning parameters related to the radiotherapy equipment, which include the machine parameters, radiation field information, and corresponding target dose distribution information required to execute the plan;

[0126] Inputting machine parameters, field information, and at least one target position information into a trained dose calculation model to obtain dose calculation distribution information output by the dose calculation model, wherein the dose calculation model is constructed based on a deep learning model;

[0127] Comparing the dose calculation distribution information output by the dose calculation model with the target dose distribution information;

[0128] Determine whether the current plan parameters meet the requirements based on the comparison results.

[0129] Of course, the computer-readable storage medium provided by an embodiment of the present invention stores a computer program not limited to the method operations described above, but can also execute related operations performed by the dose calculation system processor provided by any embodiment of the present invention, and / or execute related operations performed by the quality assurance system processor provided by any embodiment of the present invention.

[0130] The computer storage medium of the embodiment of the present invention may adopt any combination of one or more computer-readable media. The computer-readable medium may be a computer-readable signal medium or a computer-readable storage medium. The computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or component, or any combination thereof. More specific examples (a non-exhaustive list) of computer-readable storage media include: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof. In this document, a computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in combination with an instruction execution system, device or device.

[0131] A computer-readable signal medium may include a data signal propagated in baseband or as part of a carrier wave, which carries computer-readable program code. Such propagated data signals may take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A computer-readable signal medium may also be any computer-readable medium other than a computer-readable storage medium that can transmit, propagate, or transport a program for use by or in conjunction with an instruction execution system, apparatus, or device.

[0132] Program code embodied on a computer readable medium may be transmitted using any appropriate medium, including but not limited to wireless, wireline, optical fiber cable, RF, etc., or any suitable combination of the foregoing.

[0133] Computer program code for performing the operations of the present invention may be written in one or more programming languages, or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, C++, and conventional procedural programming languages ​​such as "C" or similar programming languages. The program code may be executed entirely on the user's computer, partially on the user's computer, as a stand-alone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving a remote computer, the remote computer may be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., through the Internet using an Internet service provider).

[0134] Note that the above are only preferred embodiments of the present invention and the technical principles employed. Those skilled in the art will understand that the present invention is not limited to the specific embodiments described herein, and that various obvious changes, readjustments, and substitutions can be made by those skilled in the art without departing from the scope of protection of the present invention. Therefore, although the present invention has been described in detail through the above embodiments, the present invention is not limited to the above embodiments and may include many other equivalent embodiments without departing from the concept of the present invention. The scope of the present invention is determined by the scope of the appended claims.

Claims

1. A dose calculation system, characterized in that: The method comprises a processor configured to process the following steps: Obtain dose-related parameters related to radiotherapy equipment; Inputting the dose-related parameters into a trained dose calculation model to obtain output information of the dose calculation model, wherein the dose calculation model is constructed based on a deep learning model; determining dose distribution information according to the output information; The dose distribution information is a three-dimensional dose distribution composed of dose values ​​at at least one target position; The dose-related parameters include machine parameters, field information, and at least one target position information related to the radiotherapy device; The position information is the spatial position of the water tank data, and the field information is the field size of the water tank curve.

2. The system according to claim 1, characterized in that Before inputting the dose-related parameters into the trained dose calculation model, the method further includes: Acquire training sample data, wherein the training sample data includes different sample dose parameters and sample dose information corresponding to the sample dose parameters; The pre-built dose calculation model is trained using the training sample data to obtain a trained dose calculation model.

3. The system according to claim 2, characterized in that The obtaining of training sample data comprises: A Monte Carlo algorithm is used to determine sample dose information under a plurality of sample dose parameters, and the training sample data is determined based on the plurality of sample dose parameters and the corresponding sample dose information.

4. The system according to claim 3, characterized in that The determining the training sample data based on the multiple sample dose parameters and the corresponding sample dose information includes: Some sample dose points corresponding to some of the sample dose parameters are selected as training dose points, and the training sample data are determined based on the training dose points.

5. The system according to claim 4, characterized in that Also includes: Using sample dose points other than training dose points among the sample dose points corresponding to the sample dose parameters as test dose points, and determining test sample data based on the test dose points; The trained dose calculation model is verified using the test sample data, and the trained dose calculation model is obtained based on the verification result.

6. The system according to claim 5, characterized in that The step of using the test sample data to verify the trained dose calculation model and obtaining the trained dose calculation model based on the verification result includes: Inputting the test sample data into the dose calculation model to obtain the test prediction result output by the dose calculation model; The difference between the test prediction result and the test sample result corresponding to the test sample data is calculated, and when the difference is within a set error range, the trained dose calculation model is obtained.

7. The system according to claim 3, characterized in that The sample dosage information is dosage information in water determined by using a Monte Carlo algorithm.

8. A dosage quality assurance system, characterized in that: The method comprises a processor configured to process the following steps: Acquiring planning parameters related to the radiotherapy equipment, wherein the planning parameters include machine parameters, radiation field information, and corresponding target dose distribution information required for executing the plan; Inputting the machine parameters, the field information and at least one target position information into a trained dose calculation model to obtain dose calculation distribution information output by the dose calculation model, wherein the dose calculation model is constructed based on a deep learning model; comparing the dose calculation distribution information output by the dose calculation model with the target dose distribution information; Determine whether the current plan parameters meet the requirements according to the comparison result; The dose distribution information is a three-dimensional dose distribution composed of dose values ​​at at least one target position; The position information is the spatial position of the water tank data, and the field information is the field size of the water tank curve.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by the processor, the following steps are implemented: Obtain dose-related parameters related to radiotherapy equipment; Inputting the dose-related parameters into a trained dose calculation model to obtain output information of the dose calculation model, wherein the dose calculation model is constructed based on a deep learning model; determining dose distribution information according to the output information; The dose distribution information is a three-dimensional dose distribution composed of dose values ​​at at least one target position; The dose-related parameters include machine parameters, field information, and at least one target position information related to the radiotherapy device; The position information is the spatial position of the water tank data, and the field information is the field size of the water tank curve.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by the processor, the following steps are implemented: Acquiring planning parameters related to the radiotherapy equipment, wherein the planning parameters include machine parameters, radiation field information, and corresponding target dose distribution information required for executing the plan; Inputting the machine parameters, the field information and at least one target position information into a trained dose calculation model to obtain dose calculation distribution information output by the dose calculation model, wherein the dose calculation model is constructed based on a deep learning model; comparing the dose calculation distribution information output by the dose calculation model with the target dose distribution information; Determine whether the current plan parameters meet the requirements according to the comparison result; The dose distribution information is a three-dimensional dose distribution composed of dose values ​​at at least one target position; The position information is the spatial position of the water tank data, and the field information is the field size of the water tank curve.

Images