Radiotherapy dose calculation method and device based on V-Net network

Abstract

The invention relates to the technical field of radiation dose calculation, in particular to a radiation dose calculation method and device based on a V-Net network. The method comprises the following steps: acquiring a CT image and light beam flux information for a focus area of a patient to be treated, wherein the light beam flux information represents parameter information of a light beam for performing radiotherapy on the patient to be treated; and on the basis of the CT image and the beam flux information, through a pre-trained V-Net network, performing radiotherapy dose calculation for a focus area of a patient to be treated, and generating a dose distribution diagram. According to the method, the dose distribution diagram can be automatically generated, human intervention is reduced, and the clinical working efficiency and the adaptability of a treatment plan are improved. Moreover, the pre-trained V-Net network can effectively capture features in the CT image, the calculation speed is greatly improved, the precision of the generated dose distribution diagram is equivalent to the calculation precision of the Monte Carlo method, and both the speed and the precision are taken into account.

Description

Technical Field

[0001] The present invention relates to the technical field of radiation dose calculation, and in particular to a radiotherapy dose calculation method and device based on a V-Net network. Background Art

[0002] The key to radiotherapy is to accurately calculate the dose distribution in the lesion area to ensure the effectiveness and safety of the treatment. The existing dose calculation methods mainly include:

[0003] Methods based on empirical models: such as Pencil Beam Convolution (PBC) and Anisotropic Analytical Algorithm (AAA). These methods are fast in calculation, but have limited accuracy in complex anatomical structures and in the presence of tissue heterogeneity, thus affecting the treatment effect.

[0004] Dose calculation based on the Monte Carlo method (MC): The Monte Carlo method is considered the "gold standard" for dose calculation in radiotherapy. It simulates the transmission process of particles in tissues through a large number of random sampling and can accurately calculate the dose distribution of photons and electron beams. However, due to the complex calculation process and long calculation time, it is usually only used for research or special clinical cases and is difficult to be widely used in daily clinical practice.

[0005] In summary, the calculation of radiotherapy dose is particularly important in radiotherapy planning and dynamic adaptive treatment, but the existing radiotherapy dose calculation methods are difficult to meet the needs of real-time applications. Summary of the invention

[0006] The technical problem to be solved by the present invention is to provide a radiotherapy dose calculation method and device based on a V-Net network.

[0007] The technical solution of the present invention to solve the above-mentioned technical problems is as follows: a radiotherapy dose calculation method based on a V-Net network, comprising: obtaining a CT image and beam flux information for a lesion area of ​​a patient to be treated, wherein the beam flux information represents parameter information of a beam used for radiotherapy of the patient to be treated; based on the CT image and the beam flux information, performing radiotherapy dose calculation for the lesion area of ​​the patient to be treated through a pre-trained V-Net network to generate a dose distribution map.

[0008] The beneficial effect of the present invention is that the present invention uses a pre-trained V-Net network to process CT images and beam flux information to generate a dose distribution map. The method can automatically generate a dose distribution map, reduce human intervention, and improve clinical work efficiency and adaptability of treatment plans. In addition, the pre-trained V-Net network can effectively capture the features in the CT image, greatly improve the calculation speed, and the accuracy of the generated dose distribution map is comparable to the calculation accuracy of the Monte Carlo method, achieving a balance between speed and accuracy.

[0009] Based on the above technical solution, the present invention can also be improved as follows.

[0010] Furthermore, based on the CT image and the beam flux information, the radiotherapy dose for the lesion area of ​​the patient to be treated is calculated through a pre-trained V-Net network to generate a dose distribution map, including: pre-processing the CT image to generate a numerical normalization matrix; based on the beam flux information, generating a beam distribution map; inputting the numerical normalization matrix and the beam distribution map into the pre-trained V-Net network to calculate the radiotherapy dose for the lesion area of ​​the patient to be treated, and generating a dose distribution map.

[0011] Furthermore, the preprocessing of the CT image to generate a numerical normalized matrix includes: normalizing the CT image to obtain normalized data; converting the normalized data into a four-dimensional tensor format to obtain a numerical normalized matrix.

[0012] The beneficial effect of adopting the above further scheme is: preprocessing the CT image to meet the input requirements of the V-Net network model.

[0013] Furthermore, after generating the beam distribution diagram, the method further includes: performing registration processing on the beam distribution diagram and the CT image.

[0014] The beneficial effect of adopting the above further solution is: by registering the beam distribution map with the CT image, the V-Net network model can process the beam distribution map and the CT image at the same time.

[0015] Furthermore, the pre-trained V-Net network includes an input layer, an encoder, a decoder and an output layer connected in sequence; the inputting of the numerical normalization matrix and the beam distribution map into the pre-trained V-Net network to calculate the radiotherapy dose for the lesion area of ​​the patient to be treated and generate a dose distribution map includes: inputting the numerical normalization matrix and the beam distribution map into the input layer; based on the numerical normalization matrix and the beam distribution map, performing feature extraction through the encoder to generate a feature map; based on the feature map, constructing the dose distribution map through the decoder; and outputting the dose distribution map through the output layer.

[0016] The beneficial effect of adopting the above further scheme is that the design of the V-Net network enables it to effectively capture the multi-scale features in CT images, thereby providing accurate dose calculation in the presence of complex anatomical structures and tissue heterogeneity.

[0017] Furthermore, the pre-trained V-Net network is obtained in the following manner: obtaining a training data set, the training data set including training data consisting of a CT image of the patient's lesion area, beam flux information and an actual dose distribution map, the actual dose distribution map representing the radiotherapy dose distribution for the patient's lesion area calculated using the Monte Carlo method; training a preset V-Net network based on the training data set to obtain the pre-trained V-Net network.

[0018] The beneficial effect of adopting the above further scheme is that the V-Net network is trained with the radiotherapy dose distribution for the patient's lesion area calculated by the Monte Carlo method, which can ensure that the trained V-Net network has sufficient accuracy and can maintain high accuracy close to the Monte Carlo method in complex clinical scenarios.

[0019] In order to solve the above technical problems, the present invention also provides a radiotherapy dose calculation device based on a V-Net network, comprising:

[0020] A data acquisition module, used to acquire a CT image and beam flux information of a lesion area of ​​a patient to be treated, wherein the beam flux information represents parameter information of a beam used for radiotherapy of the patient to be treated;

[0021] The dose calculation module is used to calculate the radiotherapy dose for the lesion area of ​​the patient to be treated based on the CT image and the beam flux information through a pre-trained V-Net network to generate a dose distribution map.

[0022] Further, the dose calculation module includes:

[0023] A first preprocessing unit, used for preprocessing the CT image to generate a numerical normalization matrix;

[0024] A second preprocessing unit, configured to generate a beam distribution diagram based on the beam flux information;

[0025] The dose calculation unit is used to input the numerical normalization matrix and the beam distribution map into a pre-trained V-Net network to calculate the radiotherapy dose for the lesion area of ​​the patient to be treated, and generate a dose distribution map.

[0026] To solve the above technical problems, the present invention also provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, a radiotherapy dose calculation method based on a V-Net network as described above is implemented.

[0027] To solve the above technical problems, the present invention also provides a non-transitory computer-readable storage medium, which stores computer instructions, and the computer instructions are used to enable a computer to execute a radiotherapy dose calculation method based on a V-Net network as described above. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] Figure 1 A schematic diagram of a radiotherapy dose calculation method based on a V-Net network according to the present invention;

[0029] Figure 2 A V-Net network schematic diagram of a radiotherapy dose calculation method based on a V-Net network according to the present invention;

[0030] Figure 3 A V-Net network training flow chart of a radiotherapy dose calculation method based on a V-Net network according to the present invention;

[0031] Figure 4 This is an application flow chart of a radiotherapy dose calculation method based on a V-Net network of the present invention;

[0032] Figure 5 This is a schematic diagram of a radiotherapy dose calculation device based on a V-Net network according to the present invention. DETAILED DESCRIPTION

[0033] The principles and features of the present invention are described below. The examples given are only used to explain the present invention and are not used to limit the scope of the present invention.

[0034] Embodiment 1

[0035] like Figure 1 As shown, this embodiment provides a radiotherapy dose calculation method based on a V-Net network, including:

[0036] S1. Acquire a CT image and beam flux information of a lesion area of ​​a patient to be treated, wherein the beam flux information represents parameter information of a beam used for radiotherapy of the patient to be treated.

[0037] The beam flux information can be directly derived from the Treatment Planning System (TPS). The beam flux information includes geometric information and physical information of the beam used for radiotherapy, such as the energy, direction, intensity and shape of the beam.

[0038] Specific: 1. The role of TPS: TPS is used to develop radiotherapy plans, including determining the target area, selecting the irradiation field, designing the dose distribution, etc. 2. The steps for the radiotherapy physicist to develop a radiotherapy plan in TPS: input patient information, upload CT images, outline the target area and organs at risk, select radiotherapy technology and equipment, set irradiation field parameters, etc. 3. Source of beam flux information: TPS will calculate the distribution of the beam in space based on the irradiation field parameters and generate a beam flux diagram or parameter matrix. 4. Export format: The beam flux information can be exported in the form of a file, such as DICOM format, to facilitate subsequent processing.

[0039] S2. Based on the CT image and the beam flux information, the radiotherapy dose calculation for the lesion area of ​​the patient to be treated is performed through a pre-trained V-Net network to generate a dose distribution map.

[0040] Optionally, in an embodiment, the radiotherapy dose calculation for the lesion area of ​​the patient to be treated is performed based on the CT image and the beam flux information through a pre-trained V-Net network to generate a dose distribution map, including:

[0041] S11, preprocessing the CT image to generate a numerical normalization matrix.

[0042] Optionally, in an embodiment, the preprocessing of the CT image to generate a numerical normalized matrix includes: normalizing the CT image to obtain normalized data; converting the normalized data into a four-dimensional tensor format to obtain a numerical normalized matrix.

[0043] In the process of preprocessing CT images, in order to meet the input requirements of the model, the pixel value (HU value) of the CT image needs to be normalized to between 0 and 1, and the format is converted into a four-dimensional tensor format (batch size, depth, height, width) suitable for model processing, that is, the format input to the model is (Batch, Depth, Height, Width).

[0044] S12. Generate a beam distribution diagram based on the beam flux information.

[0045] The beam flux information is converted into a beam distribution diagram, and the beam flux information can also be converted into a beam parameter matrix for subsequent processing and calculation.

[0046] Optionally, in an embodiment, after generating the beam distribution diagram, the method further includes: registering the beam distribution diagram with the CT image.

[0047] The beam distribution map is registered with the CT image so that the V-Net network model can process the beam distribution map and the CT image at the same time.

[0048] S13, inputting the numerical normalization matrix and the beam distribution map into a pre-trained V-Net network to calculate the radiotherapy dose for the lesion area of ​​the patient to be treated, and generating a dose distribution map.

[0049] Optionally, in an embodiment, Figure 2 As shown, the pre-trained V-Net network includes an input layer, an encoder, a decoder and an output layer connected in sequence; the numerical normalization matrix and the beam distribution map are input into the pre-trained V-Net network to calculate the radiotherapy dose for the lesion area of ​​the patient to be treated and generate a dose distribution map, including: inputting the numerical normalization matrix and the beam distribution map into the input layer; based on the numerical normalization matrix and the beam distribution map, feature extraction is performed through the encoder to generate a feature map; based on the feature map, the dose distribution map is constructed through the decoder; and the dose distribution map is output through the output layer.

[0050] Specifically, the encoder includes several 3D convolutional layers, followed by batch normalization layers and maximum pooling layers. Through the encoder, multi-scale features are gradually extracted. The decoder includes several deconvolutional layers (up-sampling layers), which are used to restore the low-resolution feature map extracted by the encoder to the original resolution, and at the same time, the encoder's jump connection features are combined for refinement to reconstruct the dose distribution map. The decoder uses jump connections, specifically residual connections, which enable features to be transmitted across layers in the decoder part, prevent gradient disappearance, and maintain feature information. The pre-trained V-Net network uses the Dice loss function and mean-square error (MSE) for joint optimization. The Dice loss function is suitable for segmentation tasks, and the mean-square error is suitable for regression accuracy. The design of the V-Net network enables it to effectively capture multi-scale features in CT images, thereby providing accurate dose calculation in the presence of complex anatomical structures and tissue heterogeneity.

[0051] Optionally, in an embodiment, Figure 3As shown, the pre-trained V-Net network is obtained in the following manner: obtaining a training data set, wherein the training data set includes training data consisting of a CT image of a patient's lesion area, beam flux information, and an actual dose distribution map, wherein the actual dose distribution map represents the radiotherapy dose distribution for the patient's lesion area calculated using the Monte Carlo method; and training a preset V-Net network based on the training data set to obtain the pre-trained V-Net network.

[0052] Training the V-Net network with the radiotherapy dose distribution for the patient's lesion area calculated by the Monte Carlo method can ensure that the trained V-Net network has sufficient accuracy and can maintain high accuracy close to the Monte Carlo method in complex clinical scenarios.

[0053] When training and optimizing the V-Net network model, the training data in the training data set is divided according to a certain ratio to obtain a training set, a test set, and a validation set. For example, the training set, the test set, and the validation set can be divided according to the ratio of 0.7, 0.15, and 0.15. The preset V-Net network is supervised by the training set. During the training process, the mean square error and the structural similarity index (SSIM) are used as loss functions to optimize the accuracy of the dose distribution output by the V-Net network. MSE can calculate the sum of squares of the difference between the generated dose distribution map and the true dose distribution map. SSIM is used to measure the similarity between the structure of the generated image and the true image in order to retain the local structural features of the image, thereby improving the accuracy and consistency of the dose distribution. Then, the test set and the validation set are respectively input into the V-Net network for model testing and verification. The V-Net network is fully trained to ensure that high-precision dose distribution can still be output under different anatomical structures and tissue heterogeneity conditions.

[0054] like Figure 4 As shown in the figure, in actual application, CT images of the lesion area of ​​the patient to be treated are first collected by CT equipment, and then a professional radiotherapist formulates a treatment plan for the patient to be treated in the radiotherapy planning system, and the beam flux information of the lesion area of ​​the patient to be treated is derived from the radiotherapy planning system. Through this method, based on the CT image and beam flux information, a dose distribution map is generated through a pre-trained V-Net network. Finally, the dose distribution map is applied in clinical practice to provide data reference for medical personnel.

[0055] The Monte Carlo method can also be used to calculate the radiotherapy dose distribution. After the dose distribution map is generated by the V-Net network, the dose distribution map generated by the V-Net network is compared with the radiotherapy dose distribution calculated by the Monte Carlo method to achieve more accurate radiotherapy dose calculation, and the V-Net network can be continuously optimized based on this.

[0056] In summary, the present invention uses a pre-trained V-Net network to process CT images and beam flux information to generate a dose distribution map. This method can automatically generate a dose distribution map, reduce human intervention, and improve clinical work efficiency and the adaptability of treatment plans. In addition, the pre-trained V-Net network can effectively capture the features in the CT image, greatly improve the calculation speed, and the accuracy of the generated dose distribution map is comparable to the calculation accuracy of the Monte Carlo method, achieving a balance between speed and accuracy. This method can not only meet the real-time needs in clinical applications, but also provide accurate dose distribution maps under complex anatomical structures.

[0057] This method can be seamlessly integrated into the existing TPS without modifying the software and hardware of the TPS. Only by developing a plug-in or an interface, the connection between TPS and this method can be achieved, and the supporting files of CT images and beam flux information can be directly exported from TPS for processing by this method. In addition, the present invention has a guiding role in the formulation of radiotherapy plans. The general radiotherapy plan formulation process is that the radiotherapy physicist evaluates the safety, effectiveness and feasibility of the radiotherapy plan based on the dose distribution map. If the dose distribution does not meet the treatment goal, it is necessary to adjust the irradiation field parameters, such as the shape, size, direction and angle of the irradiation field, and recalculate. Repeat the MC dose calculation until a satisfactory dose distribution is obtained. However, the MC dose calculation takes a lot of time, and repeated MC calculations require even experienced physicists to spend several hours waiting for repeated calculations. The present invention can obtain the dose distribution results in real time according to the input, reduce the time spent on dose calculation, and can serve as a basis for physicists to evaluate radiotherapy plans and guide the rapid modification of radiotherapy plans.

[0058] Embodiment 2

[0059] like Figure 5 As shown, this embodiment provides a radiotherapy dose calculation device based on a V-Net network, comprising:

[0060] A data acquisition module, used to acquire a CT image and beam flux information of a lesion area of ​​a patient to be treated, wherein the beam flux information represents parameter information of a beam used for radiotherapy of the patient to be treated;

[0061] The dose calculation module is used to calculate the radiotherapy dose for the lesion area of ​​the patient to be treated based on the CT image and the beam flux information through a pre-trained V-Net network to generate a dose distribution map.

[0062] Optionally, in an embodiment, the dose calculation module includes:

[0063] A first preprocessing unit, used for preprocessing the CT image to generate a numerical normalization matrix;

[0064] A second preprocessing unit, configured to generate a beam distribution diagram based on the beam flux information;

[0065] The dose calculation unit is used to input the numerical normalization matrix and the beam distribution map into a pre-trained V-Net network to calculate the radiotherapy dose for the lesion area of ​​the patient to be treated, and generate a dose distribution map.

[0066] Optionally, in an embodiment, the first preprocessing unit includes:

[0067] A normalization subunit, used for performing normalization processing on the CT image to obtain normalized data;

[0068] The format conversion subunit is used to convert the normalized data into a four-dimensional tensor format to obtain a numerical normalized matrix.

[0069] Optionally, in the embodiment, after the second preprocessing unit, the method further includes:

[0070] A registration unit is used to perform registration processing on the beam distribution diagram and the CT image.

[0071] Optionally, in an embodiment, the pre-trained V-Net network includes an input layer, an encoder, a decoder and an output layer connected in sequence; the inputting the numerical normalization matrix and the beam distribution map into the pre-trained V-Net network to calculate the radiotherapy dose for the lesion area of ​​the patient to be treated and generate a dose distribution map includes: inputting the numerical normalization matrix and the beam distribution map into the input layer; based on the numerical normalization matrix and the beam distribution map, performing feature extraction through the encoder to generate a feature map; based on the feature map, constructing the dose distribution map through the decoder; and outputting the dose distribution map through the output layer.

[0072] Optionally, in an embodiment, the pre-trained V-Net network is obtained in the following manner: obtaining a training data set, the training data set including training data consisting of a CT image of the patient's lesion area, beam flux information and an actual dose distribution map, the actual dose distribution map representing the radiotherapy dose distribution for the patient's lesion area calculated using the Monte Carlo method; training a preset V-Net network based on the training data set to obtain the pre-trained V-Net network.

[0073] Embodiment 3

[0074] This embodiment provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, a radiotherapy dose calculation method based on a V-Net network as described in Embodiment 1 is implemented.

[0075] Embodiment 4

[0076] This embodiment provides a non-transitory computer-readable storage medium, which stores computer instructions, and the computer instructions are used to enable a computer to execute a radiotherapy dose calculation method based on a V-Net network described in Example 1.

[0077] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner. In addition, those skilled in the art may combine and combine the different embodiments or examples described in this specification and the features of the different embodiments or examples, without contradiction.

[0078] Although the embodiments of the present invention have been shown and described above, it is to be understood that the above embodiments are exemplary and are not to be construed as limitations of the present invention. A person skilled in the art may change, modify, replace and vary the above embodiments within the scope of the present invention.

Claims

1. A radiotherapy dose calculation method based on V-Net network, characterized in that: include: Acquire a CT image and beam flux information of a lesion area of ​​a patient to be treated, wherein the beam flux information represents parameter information of a beam used for performing radiotherapy on the patient to be treated; Based on the CT image and the beam flux information, the radiotherapy dose calculation for the lesion area of ​​the patient to be treated is performed through a pre-trained V-Net network to generate a dose distribution map.

2. A radiotherapy dose calculation method based on V-Net network according to claim 1, characterized in that: The method of calculating the radiotherapy dose for the lesion area of ​​the patient to be treated based on the CT image and the beam flux information through a pre-trained V-Net network to generate a dose distribution map includes: Preprocessing the CT image to generate a numerical normalization matrix; Based on the beam flux information, generating a beam distribution diagram; The numerical normalization matrix and the beam distribution map are input into a pre-trained V-Net network to calculate the radiotherapy dose for the lesion area of ​​the patient to be treated, and generate a dose distribution map.

3. A radiotherapy dose calculation method based on V-Net network according to claim 2, characterized in that: The preprocessing of the CT image to generate a numerical normalization matrix includes: Performing normalization processing on the CT image to obtain normalized data; The normalized data is converted into a four-dimensional tensor format to obtain a numerical normalized matrix.

4. A radiotherapy dose calculation method based on V-Net network according to claim 2, characterized in that: After the beam distribution diagram is generated, the method further includes: performing registration processing on the beam distribution diagram and the CT image.

5. A radiotherapy dose calculation method based on V-Net network according to claim 2, characterized in that: The pre-trained V-Net network includes an input layer, an encoder, a decoder, and an output layer connected in sequence; the numerical normalization matrix and the beam distribution map are input into the pre-trained V-Net network to calculate the radiotherapy dose for the lesion area of ​​the patient to be treated, and generate a dose distribution map, including: Inputting the numerical normalization matrix and the beam distribution diagram into the input layer; Based on the numerical normalization matrix and the beam distribution diagram, feature extraction is performed by the encoder to generate a feature diagram; Based on the characteristic graph, constructing the dose distribution map through the decoder; The dose distribution map is output through the output layer.

6. A radiotherapy dose calculation method based on V-Net network according to claim 1, characterized in that: The pre-trained V-Net network is obtained in the following way: Acquire a training data set, wherein the training data set includes training data consisting of a CT image of a lesion area of ​​the patient, beam flux information, and an actual dose distribution map, wherein the actual dose distribution map represents a radiotherapy dose distribution for the lesion area of ​​the patient calculated using a Monte Carlo method; The preset V-Net network is trained based on the training data set to obtain the pre-trained V-Net network.

7. A radiotherapy dose calculation device based on V-Net network, characterized in that: include: A data acquisition module, used to acquire a CT image and beam flux information of a lesion area of ​​a patient to be treated, wherein the beam flux information represents parameter information of a beam used for radiotherapy of the patient to be treated; The dose calculation module is used to calculate the radiotherapy dose for the lesion area of ​​the patient to be treated based on the CT image and the beam flux information through a pre-trained V-Net network to generate a dose distribution map.

8. A radiotherapy dose calculation device based on a V-Net network according to claim 7, characterized in that: The dose calculation module comprises: A first preprocessing unit, used for preprocessing the CT image to generate a numerical normalization matrix; A second preprocessing unit, configured to generate a beam distribution diagram based on the beam flux information; The dose calculation unit is used to input the numerical normalization matrix and the beam distribution map into a pre-trained V-Net network to calculate the radiotherapy dose for the lesion area of ​​the patient to be treated, and generate a dose distribution map.

9. An electronic device, characterized in that: It includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, a radiotherapy dose calculation method based on a V-Net network as described in any one of claims 1 to 6 is implemented.

10. A non-transitory computer-readable storage medium storing computer instructions, characterized in that: The computer instructions are used to enable a computer to execute a radiotherapy dose calculation method based on a V-Net network as described in any one of claims 1 to 6.

Images