JP7412083B2 - Treatment planning support device and method - Google Patents

Description

Embodiments of the present invention relate to a treatment planning support device and method.

Radiation therapy (RT) is generally a treatment that is completed until the end based on the treatment plan decided at the beginning. Therefore, the quality of the treatment plan initially determined influences the therapeutic effect and occurrence of side effects after RT. With the development of information technology, it has become possible to obtain prediction results of treatment effects and the probability of occurrence of side effects after RT based on image information obtained from CT images at the time of treatment planning and information such as dose. However, in recent years, with the development of technology that facilitates treatment planning, a treatment method (hereinafter referred to as Adaptive RT) that revises the treatment plan based on the therapeutic effects and occurrence of side effects during the treatment period has been attracting attention. Since the above prediction results are intended to support decision-making regarding the quality of the treatment plan created at the time of treatment planning, it is difficult to determine whether to apply Adaptive RT based on the therapeutic effects or occurrence of side effects during the treatment period.

Zhi Cheng, et al, “Evaluation of classification and regression tree (CART) model in weight loss prediction following head and neck cancer radiation therapy”, Advances in Radiation Oncology (2018), December 7, 2017, volume 3, issue 3, 346 -355

The problem to be solved by the present invention is to be able to easily and accurately support the decision to apply Adaptive RT during the treatment period.

The treatment planning support device according to this embodiment includes an acquisition unit and a generation unit. The acquisition unit acquires information regarding a first treatment plan related to radiation therapy. The generation unit generates information regarding a second treatment plan using information regarding the first treatment plan and a medical image acquired when performing radiation therapy based on the first treatment plan.

FIG. 1 is a block diagram of a treatment planning support device according to a first embodiment. FIG. 2 is a diagram showing a concept during learning of a machine learning model used in the treatment planning support device according to the first embodiment. FIG. 3 is a diagram showing a concept when using a machine learning model used in the treatment planning support device according to the first embodiment. FIG. 4 is a flowchart showing the operation of the treatment planning support device according to the first embodiment. FIG. 5 is a flowchart showing a modified example of the operation of the treatment planning support device according to the first embodiment. FIG. 6 is a diagram showing a first presentation example of prognosis prediction information according to the first embodiment. FIG. 7 is a diagram showing a second presentation example of prognosis prediction information according to the first embodiment. FIG. 8 is a block diagram of a treatment planning device according to the second embodiment, which includes the treatment planning support device according to the first embodiment.

Hereinafter, a treatment planning support device and method according to the present embodiment will be explained with reference to the drawings. In the following embodiments, parts with the same reference numerals perform similar operations, and redundant explanations will be omitted as appropriate. Hereinafter, one embodiment will be described using the drawings.

(First embodiment)
FIG. 1 is a block diagram showing the configuration of a treatment planning support device 1 according to the first embodiment. The treatment planning support device 1 includes a processing circuit 11 and a memory 13. The processing circuit 11 is composed of processors such as a CPU (Central Processing Unit) and a GPU (Graphics Processing Unit) as hardware resources. The processing circuit 11 includes an acquisition function 111 , a generation function 113 , a prediction function 115 , a learning function 117 , and a presentation function 119 . Note that the processing circuit 11 may be realized by an ASIC (Application Specific Integrated Circuit), an FPGA (Field Programmable Gate Array), a CPLD (Complex Programmable Logic Device), an SPLD (Simple Programmable Logic Device), etc. that can realize the above functions. good.

The acquisition function 111 acquires information regarding a first treatment plan related to radiation therapy. The first treatment plan may be an initially established treatment plan or a treatment plan modified by Adaptive RT.

The generation function 113 generates information regarding an assumed second treatment plan using information regarding the first treatment plan and medical images acquired when performing radiation therapy based on the first treatment plan. do. As the medical image, an image generated by a cone beam CT device, an X-ray computed tomography device, a magnetic resonance imaging device, a nuclear medicine diagnostic device, or the like may be used. Hereinafter, the medical image used by the generation function 113 is a cone beam CT image (hereinafter also referred to as a CBCT image) or a CT image generated by a CT imaging device such as a cone beam CT device or an X-ray computed tomography device. do. The generation function 113 may calculate image feature amounts from the CT image and the CBCT image. A general method may be used to calculate the image feature amount. Further, a specific example of the image feature amount will be described later. The generation function 113 may also design a first treatment plan that includes a treatment plan that is initially created.

The prediction function 115 uses first prognosis prediction information, which is a prediction of the prognosis expected when treatment is proceeded according to the first treatment plan, and treatment according to the second treatment plan generated by the generation function 113. second prognosis prediction information that is a prediction of the prognosis expected when the process is advanced. Prognosis prediction includes, for example, tumor response that predicts the degree of shrinkage after radiation irradiation and the presence or absence of side effects. The tumor response is indicated by, for example, the probability of complete cure of the tumor or the degree of reduction from the size before treatment. Note that if the tumor is larger than before treatment, it may be expressed as a negative value. The presence or absence of side effects is indicated by, for example, the presence or absence of weight loss.

The learning function 117 trains a model for generating the first prognosis prediction information and the second prognosis prediction information. Here, we assume that a trained model is generated by learning a machine learning model such as a neural network. Note that the model generated by the learning function 117 is not limited to machine learning, and the model generated by the learning function 117 may be a look-up table (LUT).

The presentation function 119 presents the first prognosis prediction information and the second prognosis prediction information to the user. As a presentation method, the first prognosis prediction information and the second prognosis prediction information may be displayed on a display (not shown), or the first prognosis prediction information and the second prognosis prediction information may be displayed on a speaker (not shown). (not shown) may be output (played) as audio.

The memory 13 is configured of hardware resources such as ROM (Read Only Memory), RAM (Random Access Memory), HDD (Hard Disk Drive), and SSD (Solid State Drive). The memory 13 is assumed to store information regarding the first treatment plan, a machine learning model, and learning data. Further, the first prognosis prediction information and the second prognosis prediction information calculated by the prediction function 115 may be stored. The processing circuit 11 and the memory 13 are communicably connected to each other via a bus.

Next, the concept of learning the machine learning model used in the treatment planning support device 1 will be explained with reference to FIG. 2.
FIG. 2 shows input of learning data to the machine learning model 210 before learning. The learning data is a set of various information obtained at the time of the first treatment plan and information on treatment effects and occurrence of side effects after the first treatment is completed.

The machine learning model 210 receives as input data the treatment plan information 201, the CT image feature amount 203, and the patient-related information 205 at the time of the first treatment plan. Items included in the treatment plan information 201 include, for example, treatment site, radiation dose distribution, irradiation method such as radiation irradiation angle and number of irradiation, and radiation treatment conditions. The CT image feature amount 203 is assumed to be numerical data such as the brightness value or CT value, shape, texture, and statistical amount for the CT image used when creating the first treatment plan. Note that the CT image is a multi-slice CT image that is taken before the start of radiation therapy when creating a treatment plan. The patient-related information 205 includes, for example, the patient's age, sex, treatment site, body shape information, cancer stage information, medical history, and cases of the patient's family.

On the other hand, the correct data used for learning the machine learning model 210 is the first prognosis information 207. The first prognostic information 207 includes the above-mentioned prognostic information based on the first treatment plan, that is, information regarding side effects, such as whether the patient has experienced side effects, and whether the size of the tumor has decreased. , including information on treatment effects, such as whether the status quo is maintained or whether there is a relapse. As the first prognosis information 207, prognosis information obtained by performing radiotherapy on a patient based on the first treatment plan and observing the patient's progress may be used.

The memory 13 stores learning data generated from an example of a treatment plan for a certain patient. In other words, the memory 13 uses treatment plan information 201, CT image features 203, and patient-related information 205 as input data when performing radiotherapy on a patient based on the first treatment plan, and uses prognosis information 207 regarding the patient as correct answer data. It is sufficient to store one set of learning data.

Using the learning function 117, the processing circuit 11 performs machine learning on the machine learning model 210 using the learning data according to the model learning program, and generates a learned model. The machine learning model 210 is assumed to be a neural network, deep learning, random forest, support vector machine, kernel regression, etc., but is not limited thereto and may be any other machine learning model.

Note that the trained model may be generated by, for example, learning the machine learning model 210 based on learning data prepared in advance at the time of factory shipment, or may be generated in an untrained state at the time of factory shipment and sent to the shipping destination (e.g. Hospital), the treatment planning support device 1 may acquire the learning data by connecting to a medical information server, and the machine learning model 210 may be trained by the learning function 117 based on the acquired learning data. Furthermore, the learning function 117 may be configured to update the learned model based on newly generated learning data.

Next, the concept of using the trained model used in the treatment planning support device 1 will be explained with reference to FIG. 3.
When using the trained model 307 obtained by machine learning in FIG. , patient-related information 205 at the time of the first treatment plan are input to the learned model 307.

Here, the generation function 113 causes the processing circuit 11 to assume a second treatment plan using the treatment plan information 201 at the time of the first treatment plan and the CBCT image 301 during radiation treatment based on the first treatment plan. Treatment plan information 303 is generated. Specifically, the CBCT image 301 is regarded as a CT image for treatment planning, and the image features (parameters) such as the tumor size considered in the first treatment planning information 201 and the contour information that determines the irradiation position are used as the CBCT image. 301, the treatment plan information 303 assuming the second treatment plan may be generated. To update the contour information, registration (image positioning) between the CT image and the CBCT image 301 during the first treatment planning may be used. Note that the second treatment plan is created by taking into account the treatment situation (integrated dose of radiation, etc.) according to the first treatment plan immediately before generating the second treatment plan, and calculates the remaining irradiation in the first treatment plan. Planned based on quantity.
The CBCT image feature amount 305 is generated by, for example, the generation function 113 calculating a feature amount similar to the CT image feature amount 203 from the CBCT image 301 .

In this way, by setting the value of the input data as data related to the second treatment plan information 303 without changing the type of input data on which the machine learning model is trained, the learned model 307 itself can be used as the first treatment plan information. Although it has been trained to output the prediction result of the first prognosis information 207 based on the learned model 307, as a result, the prognosis prediction information 309 (hereinafter, the second Prognosis prediction information 309) can be output.
In this embodiment, it has been explained that the second treatment plan information 303 assuming the second treatment plan is used to output the prognosis prediction information 309 in the second treatment plan. By inputting the first treatment plan information 201 to the learned model 307 instead of 303, it is possible to output prognosis prediction information when treatment according to the first treatment plan is continued.

Next, an example of the operation of the treatment planning support device 1 according to this embodiment will be described with reference to the flowchart of FIG. 4.
In step S401, a CT image for treatment planning is captured using a CT imaging device (not shown).
In step S402, a first treatment plan is created based on the CT image taken in step S401. Note that the first treatment plan may be designed by the processing circuit 11 using a generation function, or may be designed externally (for example, by a treatment planning device).
After the first treatment plan is established, the patient is treated according to the first treatment plan. Here, it is assumed that a patient visits a hospital and enters a treatment room (step Sa).
In step S403, a CBCT image before the radiation therapy is performed is captured by the CT imaging device.

In step S404, radiation therapy for that day is performed based on the first treatment plan. For example, irradiation is carried out according to a plan such as how many minutes a day a predetermined dose of irradiation is to be performed and how many times every few days. Note that the treatment plan information is transmitted to a radiation therapy apparatus (not shown). A radiation therapy device irradiates a patient's tumor with radiation according to a treatment plan.
In step S405, it is determined whether there is an instruction from the user to determine the applicability of Adaptive RT. For example, when the acquisition function 111 acquires an instruction from the user to rebuild a treatment plan for Adaptive RT, or an instruction to perform prognosis prediction to consider whether it is better to apply Adaptive RT, the Adaptive RT It may be determined that there has been an instruction to determine the applicability of RT. If there is a determination instruction from the user, the process advances to step S407, and if there is no user instruction to determine the applicability of Adaptive RT, the process advances to step S406.

In step S406, it is determined whether the radiation therapy based on the first treatment plan has been completed, that is, whether the prescribed number of radiation irradiations planned by the first treatment plan have been completed. For example, the processing circuit 11 may obtain information on the number of times of radiation irradiation with respect to the prescribed number of times, and if the prescribed number of times and the number of radiation irradiations match, it may be determined that the prescribed number of radiation irradiations have been completed. If the prescribed number of radiation irradiations have been completed, the process is terminated; if the prescribed number of radiation irradiations have not been completed, the process returns to step Sa for the next treatment and continues radiotherapy based on the first treatment plan. do.

In step S407, the prediction function 115 performs prognosis prediction in order for the processing circuit 11 to determine whether or not to apply Adaptive RT. Specifically, the prediction function 115 causes the processing circuit 11 to generate treatment plan information assuming a second treatment plan based on a CBCT image taken during radiation therapy performed immediately before the instruction from the user. generate. Using the prediction function 115, the processing circuit 11 inputs treatment plan information based on the first treatment plan, CT image features, and patient-related information to the learned model, and obtains first prognosis prediction information as an output of the learned model. . Further, the processing circuit 11 uses the prediction function 115 to input the treatment plan information based on the second treatment plan, the CBCT image feature amount and patient information regarding the CBCT image acquired in step S403 to the learned model, and outputs the learned model. and second prognosis prediction information.

In step S408, the processing circuit 11 uses the presentation function 119 to present the first prognosis prediction information and the second prognosis prediction information. The presentation of the first prognosis prediction and the second prognosis prediction is assumed to be displayed on a display, for example, but the notification is not limited to this, and may be notified by voice, or by hard copy and output on paper media. good.

In step S409, the processing circuit 11 obtains an instruction as to whether or not to apply Adaptive RT. In other words, in order to apply Adaptive RT, the processing circuit 11 uses the Adaptive It is determined that the instruction to apply the treatment plan has been obtained (the second treatment plan has been selected). If an instruction to apply Adaptive RT is acquired, the process advances to step S410, and if an instruction to apply Adaptive RT is not acquired, that is, if Adaptive RT is not applied, the process returns to step S406 and the process is repeated.
Note that the treatment planning support device 1 may automatically determine whether or not to apply Adaptive RT. For example, if the processing circuit 11 determines based on the first prognosis prediction information and the second prognosis prediction information that the second prognosis prediction information has a better result than the first prognosis prediction information (for example, If the probability of occurrence is below a threshold), it may be determined that Adaptive RT is to be applied, and the user may be presented with a recommendation to apply Adaptive RT.

In step S410, the generation function 113 causes the processing circuit 11 to switch from the first treatment plan to the second treatment plan. Thereafter, radiotherapy is performed based on the second treatment plan. Note that the second treatment plan may use treatment plan information assumed as the second treatment plan based on the CBCT image used in step S407, or a new multi-slice CT image is taken for treatment planning, A second treatment plan may be established based on the multi-slice CT image.

Although the flowchart in FIG. 4 assumes that prognosis prediction is performed using a user's instruction as a trigger, prognosis prediction may be performed automatically in the background.

An example of the operation of the treatment planning support device when prognosis prediction is performed in the background will be described with reference to the flowchart of FIG. Steps S401 to S404, S406, and S408 to S410 are the same processes as in FIG. 4, so their explanations will be omitted.

In step S501, the prediction function 115 causes the processing circuit 11 to use the CBCT image before radiation therapy in step S403 to perform a first A prognosis prediction and a second prognosis prediction are performed to obtain first prognosis prediction information and second prognosis prediction information, respectively.

In step S502, for example, the processing circuit 11 uses the prediction function 115 to determine whether to present the first prognosis prediction information and the second prognosis prediction information. For example, it may be determined that the first prognosis prediction information and the second prognosis prediction information are to be presented when the difference between the first prognosis prediction and the second prognosis prediction is greater than or equal to a threshold value. Specifically, when the prognosis prediction is the probability of occurrence of a side effect, the first prognosis prediction information and the second prognosis prediction information are present. Alternatively, if the second prognosis prediction has a more favorable result than the first prognosis prediction (e.g., the probability of occurrence of side effects is low), it is determined that the first prognosis prediction information and the second prognosis prediction information are to be presented. You can. Furthermore, when at least one of the first prognosis prediction information and the second prognosis prediction information indicates that the prognosis prediction is below a threshold, for example, the probability of occurrence of side effects is below a threshold, the first prognosis prediction It may be determined to present the information and the second prognostic information.
If it is determined that the first prognosis prediction information and the second prognosis prediction information are to be presented, the process advances to step S408. On the other hand, if the first prognosis prediction is more favorable than the second prognosis prediction, the treatment can be continued based on the first treatment plan, so the process returns to step S406 and repeats the process.

Note that the first prognosis prediction information and the second prognosis prediction information may be presented not only as a result of the prognosis prediction but also after the implementation of the prognosis prediction, or every time a predetermined number of treatments are completed (e.g. , every second time, every fifth time), the first prognosis prediction and the second prognosis prediction may be presented.

Next, a first presentation example of prognosis prediction information will be described with reference to FIG. 6.
FIG. 6 is an example of displaying the probability of occurrence of side effects in the first prognosis prediction and the probability of occurrence of side effects in the second prognosis prediction when Adaptive RT is applied. In the example of FIG. 6, the first prognosis prediction is displayed as <probability of occurrence of serious side effects>, and the probability of occurrence of serious pneumonitis 3 months after treatment is "87%". On the other hand, the second prognosis prediction is displayed as <when Adaptive RT is applied>, and the probability of occurrence of severe pneumonitis 3 months after treatment is "65%". Therefore, in this case, it can be seen that the probability of side effects occurring is lower when Adaptive RT is applied. Thereby, the user of the treatment planning support device can easily determine whether to apply Adaptive RT by referring to the two presented probabilities of occurrence.

Next, a second presentation example of prognosis prediction information will be described with reference to FIG. 7.
In FIG. 7, only the prognosis prediction with a favorable prediction result is presented from the first prognosis prediction information and the second prognosis prediction information. That is, for example, if the probability of occurrence of side effects is lower in the second prognosis prediction than in the first prognosis prediction, only the second prognosis prediction, which is a case in which Adaptive RT is applied, is presented. On the other hand, if the probability of occurrence of side effects is lower in the first prognosis prediction than in the second prognosis prediction, only the first prognosis prediction is presented.

Note that if the probability of occurrence of side effects is lower in the first prognosis prediction than in the second prognosis prediction, the processing circuit 11 does not present the prognosis prediction information because the treatment can proceed with the current treatment plan. All right. Furthermore, the processing circuit 11 may present the second prognosis prediction when the probability of occurrence of side effects is lower in the second prognosis prediction than in the first prognosis prediction, and may further prompt the user to apply Adaptive RT. A message (or voice, etc.) may be sent to the effect that it is recommended.
According to another example shown in FIG. 7, the user can decide whether to apply Adaptive RT based on one presented prognosis prediction without comparing two prognosis predictions.

According to the first embodiment described above, the prognosis based on the CBCT image taken while performing the treatment based on the first treatment plan is based on the assumption of the second treatment plan when Adaptive RT is applied. Generate predictions using a trained model. That is, it becomes possible to obtain a prognosis prediction based on the second treatment plan without capturing a CT image for treatment planning again. This allows users such as doctors to easily understand how prognosis prediction changes when Adaptive RT is applied. Further, since there is no need to take a CT image again when obtaining a prognosis prediction based on the second treatment plan, it is also possible to reduce the burden on the patient. Therefore, even if the user is not familiar with Adaptive RT, the user can accurately and easily judge whether to change the treatment plan from the presented results. As a result, a more appropriate treatment plan can be formulated and the patient's QOL can also be improved. In other words, the decision to apply Adaptive RT during the treatment period can be supported with high accuracy and ease.

(Second embodiment)
In the second embodiment, as an implementation example of the treatment planning support device according to the first embodiment, an example in which the treatment planning support device according to the first embodiment is included in the treatment planning device is shown in the block diagram of FIG. . Although the second embodiment shows an example in which a treatment planning support device is installed in a treatment planning device, the present invention is not limited to this. Good too.

As shown in FIG. 8, the treatment planning device 2 according to the second embodiment includes a processing circuit 11, an image processing circuit 50, a communication interface (IF) 60, a display 70, an input interface (IF) 80, and a memory 90. . The processing circuit 11, the image processing circuit 50, the communication interface 60, the display 70, the input interface 80, and the memory 90 are communicably connected to each other via a bus.

The processing circuit 11 realizes each function included in the treatment planning support device 1.
The patient information database 30 is a database that stores patient information of a plurality of patients. The patient information includes basic patient information and treatment result information. The patient basic information is basic information for identifying a patient, and includes, for example, patient ID, patient name, gender, and age. The treatment result information is information regarding the results of treatment actually administered to the patient.

The image processing circuit 50 includes a processor such as a CPU or a GPU, and a memory such as a ROM or RAM as hardware resources. The image processing circuit 50 performs various image processing on the treatment plan image. For example, the image processing circuit 50 performs three-dimensional image processing such as volume rendering, surface volume rendering, image value projection processing, MPR (Multi-Planer Reconstruction) processing, and CPR (Curved MPR) processing on a three-dimensional treatment plan image. to generate a two-dimensional medical image for display. Note that the image processing circuit 50 may be realized by an ASIC, FPGA, CPLD, or SPLD that can realize the above image processing.

The communication interface 60 communicates with radiation therapy equipment, a picture archiving and communication system (PACS), a hospital information system (HIS), and a radiology information system (RIS) via a wired or wireless connection (not shown). It has a communication interface that performs data communication with a radiology information system (OIS), a radiotherapy information management system (OIS), and the like.

Display 70 displays various information. Specifically, display 70 includes a display interface circuit and a display device. The display interface circuit converts data representing a display object into a video signal. The video signal is supplied to a display device. The display device displays a video signal representing a display target. As the display device, for example, a CRT display, a liquid crystal display, an organic EL display, an LED display, a plasma display, or any other display known in the art can be used as appropriate.

Specifically, the input interface 80 includes an input device and an input interface circuit. The input device receives various commands from the user. As input devices, a keyboard, mouse, various switches, etc. can be used. The input interface circuit supplies the output signal from the input device to the processing circuit 11 via the bus.

The memory 90 is a storage device such as an HDD, SSD, or integrated circuit storage device that stores various information. For example, the memory 90 stores machine learning models before learning, trained models, and learning data. As hardware, the memory 90 may be a drive device or the like that reads and writes various information to/from a portable storage medium such as a CD-ROM drive, a DVD drive, or a flash memory.

According to the second embodiment described above, by installing the treatment planning support device in the treatment planning device, how the prognosis prediction changes when Adaptive RT is applied, as in the first embodiment. more accurately and change the treatment plan if necessary.

Although several embodiments of the invention have been described, these embodiments are presented by way of example and are not intended to limit the scope of the invention. These novel embodiments can be implemented in various other forms, and various omissions, substitutions, and changes can be made without departing from the gist of the invention. These embodiments and their modifications are included within the scope and gist of the invention, as well as within the scope of the invention described in the claims and its equivalents.

1 Treatment planning support device 2 Treatment planning device 11 Processing circuit 13 Memory 30 Patient information database 50 Image processing circuit 60 Communication interface 70 Display 80 Input interface 90 Memory 111 Acquisition function 113 Generation function 115 Prediction function 117 Learning function 119 Presentation function 201,303 Treatment planning information 203 CT image feature amount 205 Patient related information 207 Prognosis information 210 Machine learning model 301 CBCT image 305 CBCT image feature amount 307 Learned model 309 Prognosis prediction information

Claims (9)

an acquisition unit that acquires information regarding a first treatment plan related to radiation therapy;
a generation unit that generates information regarding a second treatment plan using information regarding the first treatment plan and a medical image acquired when performing radiation therapy based on the first treatment plan;
According to the model that outputs the first prognosis prediction information assumed when the treatment is proceeded according to the first treatment plan, the second prognosis prediction information assumed when the treatment is proceeded according to the second treatment plan. a prediction unit that outputs prognosis prediction information;
a presentation unit that presents the first prognosis prediction information and the second prognosis prediction information when the difference between the first prognosis prediction information and the second prognosis prediction information is equal to or greater than a threshold;
Equipped with
The model uses information regarding past treatment plans, information regarding medical images from the time of the past treatment plan, and patient information of patients treated according to the past treatment plan as input data, and calculates the prognosis based on the treatment plan. Generated by learning information as correct data,
The prediction unit includes information regarding a treatment plan assuming the second treatment plan, information regarding images being treated according to the first treatment plan, and patient information of a patient being treated according to the first treatment plan. by inputting into the model, outputting the second prognosis prediction information,
The second treatment plan is planned based on the allowable remaining irradiation dose based on the treatment situation in the first treatment plan.
Treatment planning support device. The prediction unit outputs the second prognosis prediction information according to the model using information regarding the second treatment plan, patient information, and image features of the medical image.
The treatment planning support device according to claim 1. The prediction unit outputs the first prognosis prediction information and the second prognosis prediction information,
The treatment planning support device according to claim 1 or 2. The model is generated by machine learning based on information obtained during the first treatment plan and first prognosis information after treatment based on the first treatment plan.
The treatment planning support device according to any one of claims 1 to 3. The presentation unit presents the first prognosis prediction information and the second prognosis prediction information, whichever has a better prognosis prediction result.
The treatment planning support device according to any one of claims 1 to 4. When the second prognosis prediction information has a better prognosis prediction result than the first prognosis prediction information, the presentation unit changes the treatment plan to the second treatment plan during treatment according to the first treatment plan. presents a message recommending the application of Adaptive RT indicating that
The treatment planning support device according to any one of claims 1 to 5. The first prognosis prediction information and the second prognosis prediction information include information regarding the probability of occurrence of side effects or information regarding tumor response,
The treatment planning support device according to any one of claims 1 to 6. The medical image is a cone beam CT image.
The treatment planning support device according to any one of claims 1 to 7. Obtaining information regarding the first treatment plan related to radiation therapy,
Generating information regarding a second treatment plan using information regarding the first treatment plan and a medical image acquired when performing radiation therapy based on the first treatment plan;
According to the model that outputs the first prognosis prediction information assumed when the treatment is proceeded according to the first treatment plan, the second prognosis prediction information assumed when the treatment is proceeded according to the second treatment plan. Output prognosis prediction information of
presenting the first prognosis prediction information and the second prognosis prediction information when the difference between the first prognosis prediction information and the second prognosis prediction information is a threshold value or more ;
The model uses information regarding past treatment plans, information regarding medical images from the time of the past treatment plan, and patient information of patients treated according to the past treatment plan as input data, and calculates the prognosis based on the treatment plan. Generated by learning information as correct data,
Information regarding a treatment plan assuming the second treatment plan, information regarding images being treated according to the first treatment plan, and patient information of a patient being treated according to the first treatment plan are input into the model. By doing so, the second prognosis prediction information is output,
The second treatment plan is planned based on the allowable remaining irradiation dose based on the treatment situation in the first treatment plan.
Treatment planning support method.