CN107441637B - Intensity modulated radiation therapy 3-dimensional dose is distributed in the works prediction technique and its application - Google Patents
Intensity modulated radiation therapy 3-dimensional dose is distributed in the works prediction technique and its application Download PDFInfo
- Publication number
- CN107441637B CN107441637B CN201710851175.7A CN201710851175A CN107441637B CN 107441637 B CN107441637 B CN 107441637B CN 201710851175 A CN201710851175 A CN 201710851175A CN 107441637 B CN107441637 B CN 107441637B
- Authority
- CN
- China
- Prior art keywords
- dose
- patient
- voxel
- model
- radiation therapy
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Classifications
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61N—ELECTROTHERAPY; MAGNETOTHERAPY; RADIATION THERAPY; ULTRASOUND THERAPY
- A61N5/00—Radiation therapy
- A61N5/10—X-ray therapy; Gamma-ray therapy; Particle-irradiation therapy
- A61N5/103—Treatment planning systems
- A61N5/1031—Treatment planning systems using a specific method of dose optimization
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61N—ELECTROTHERAPY; MAGNETOTHERAPY; RADIATION THERAPY; ULTRASOUND THERAPY
- A61N5/00—Radiation therapy
- A61N5/10—X-ray therapy; Gamma-ray therapy; Particle-irradiation therapy
- A61N5/103—Treatment planning systems
- A61N5/1039—Treatment planning systems using functional images, e.g. PET or MRI
Landscapes
- Health & Medical Sciences (AREA)
- Engineering & Computer Science (AREA)
- Biomedical Technology (AREA)
- Pathology (AREA)
- Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
- Radiology & Medical Imaging (AREA)
- Life Sciences & Earth Sciences (AREA)
- Animal Behavior & Ethology (AREA)
- General Health & Medical Sciences (AREA)
- Public Health (AREA)
- Veterinary Medicine (AREA)
- Radiation-Therapy Devices (AREA)
Abstract
The invention discloses a kind of intensity modulated radiation therapy prediction techniques that 3-dimensional dose is distributed in the works, and steps are as follows: (1) collecting effective intensity modulated radiation therapy planning data and form case database；(2) according to the resolution sizes of CT image, the target area (PTV) of patient, each organ that jeopardizes are divided into several small voxels；(3) anatomical features of each patient in database are extracted；(4) dose characteristics of each patient in database are extracted；(5) artificial neural network is built, the anatomical features and dose characteristics of patient are inputted, goes out the mapping relations between anatomical features and dose characteristics by artificial neural network learning, obtains the correlation model of the two；(6) it is distributed using the 3-dimensional dose of the new patient of correlation model prediction.The application of the above method carries out patient and jeopardizes the dose prediction of organ, and realize quality control to use above-mentioned dosage distribution forecasting method.By the above-mentioned means, the present invention can be realized the 3-dimensional dose forecast of distribution of intensity modulated radiation therapy plan, and it can be applied to Quality Control Links.
Description
Technical field
The present invention relates to medical radiotherapy technical field, in particular to intensity modulated radiation therapy in the works 3-dimensional dose distribution it is pre-
The concrete application of survey method and this method.
Background technique
Tumour radiotherapy using ray kill tumour cell, and be avoided as much as normal tissue and jeopardize organ by
Irradiation is one of the main means for treating malignant tumour.Realization beam energy high concentration is to improve radiotherapy matter in tumor region
The key of amount.On radiotherapy technology, intensity-modulated radiation therapy technology adjusts strong technology, by the movement of multi-leaf optical grating and penetrates
The technology of beam intensity adjusting etc. further improves the conformal degree of patient's body dosage distribution, is prostate, brain neck, encephalic etc.
Mainstream therapeutic modality in the tumour of type.
Clinically, doctor can referring to radiotherapy planning clinical procedure, such as PTV coverage rate, jeopardize the tolerance dose of organ,
To assess whether a plan can be used in patient.The specification sets different plan mesh for different types of tumour
Mark, these targets are all based on the data on patient census.Use and lack patient-specific data as the target of plan, is
Lead to the one of the major reasons that clinically radiotherapy planning quality good or not is different.In addition to this, the quality of plan further depends on physics
The experience of teacher and time and energy spent by single plan above.Physics teacher rule of thumb judges radiotherapy planning system (TPS)
Whether there is room for improvement for the plan of generation, if cannot receive this plan, needs to reset injectivity optimizing target, directly
To optimal radiotherapy planning is obtained, this is the process of a continuous trial and error.Corresponding to the correct precompensation parameter variation of physics Shi Ruoneng
Results change, obtain an optimal plan for more sure.But due to the limitation of experience and time, radiotherapy planning is reaching
After clinical procedure, physics Shi Buzai is further optimized, so that patient inadvertently receives the radiotherapy of suboptimum.From
It is seen on clinical meaning, target area realizes that the maximum dose of surrounding tissue and organ is exempted under the premise of meeting prescribed dose irradiation
It is the important channel for ensureing patient's therapeutic quality.
Strong dosage in the works is adjusted to be distributed in the parameters such as the optimization aim of the prescribed dose that doctor provides and physics teacher setting
Constraint under, it is highly conformable with target region shape.Therefore, there are certain to be associated between patient anatomy and dosage, can establish
The correlation model of the two with realize dosage distribution prediction.
2011, the A planning quality evaluation that zhu et al. is delivered on Med Phys publication
Tool for prostate adaptive IMRT based on machine learning describes a kind of description dissection
The method of feature, the description describe to jeopardize the spatial relationship of organ (OAR) and target area (PTV) from distance and volume, however
This description method only considers influence of the spatial relationship of two OAR and PTV to the OAR institute acceptable dose, does not consider OAR completely
Between influence each other.In addition to this, the dose characteristics that this method is selected are the two-dimensional signals by extraction, cannot sufficiently be reflected
Performance situation on the three-dimensional space of dosage distribution.
Summary of the invention
It is an object of the invention to disclose a kind of 3-dimensional dose distribution forecasting method suitable for intensity modulated radiation therapy plan, with
Solve the problems such as anatomic information description is not comprehensive in the prior art.It is another object of the present invention to provide the prediction techniques
Concrete application.
In order to achieve the above objectives, the present invention adopts the following technical solutions realization: 3-dimensional dose is distributed intensity modulated radiation therapy in the works
Prediction technique, realized by predicting the dosage of each voxel 3-dimensional dose distribution prediction, comprising the following steps:
(1) it collects effective intensity modulated radiation therapy planning data and forms case database；
(2) according to the resolution sizes of CT image, the target area (PTV) of patient, each organ that jeopardizes are divided into several
Small voxel；
(3) anatomical features of each patient in database are extracted, including PTV volume, small voxel are to the boundary PTV most narrow spacing
From, small voxel to each minimum range for jeopardizing organ boundaries；
(4) dose characteristics of each patient in database are extracted；
(5) artificial neural network is built, the anatomical features and dose characteristics of patient are inputted, by artificial neural network learning
Mapping relations between anatomical features and dose characteristics out obtain the correlation model of the two；
(6) it is distributed using the 3-dimensional dose of the new patient of correlation model prediction.
As a preference, small voxel to PTV and each minimum range for jeopardizing organ boundaries are all in the step (3)
To PTV or jeopardize the minimum ranges of organ boundaries for voxel small on three-dimensional space.
As a preference, the anatomical features and dose characteristics of patient are in input nerve net in the step (3) and (4)
Before network, be first normalized, by characteristic value each in anatomical features and dose characteristics value distinguish Linear Mapping to [-
1,1] in range.
As a preference, the foundation of correlation model is specifically included by MATLAB software realization in the step (5)
Following steps:
(5.1) artificial neural network tool box is called in MATLAB；
(5.2) neural network is built using MATLAB artificial neural network tool box, which is three layers of nerve net
Network；Input neuronal quantity is determined by anatomical features quantity；Output neuron quantity is 1；Hidden layer neuron quantity determines
Within the scope of 3 to twice of input neuronal quantity, then the specific quantity of hidden layer neuron is determined by model verifying；
Training function selects Regularization algorithms；Excitation function selects tanh S type function；
(5.3) extracted anatomical features and dose characteristics are inputted into network, trains the correlation model of the two.
As a preference, determining that hidden layer neuron is specific amount of by model verifying in the step (5.2)
Specific steps are as follows:
(5.2.1) establishes the different neural network of hidden layer neuron quantity, net according to hidden layer neuron quantitative range
Network quantity is determined by neuronal quantity range；
Existing case is divided into training group and test group according to the ratio of 70% and 30% by (5.2.2)；
(5.2.3) sequentially inputs the anatomical features of patient each in training group and dose characteristics in single Neural,
A model is trained, in this way, going out the model of respective numbers using all neural metwork trainings；
(5.2.4) sequentially inputs the anatomical features of patient each in test group in single model, obtains each patient and exists
Predicted dose characteristic value in single model, so calculates predicted dose characteristic value of each patient in each model；
(5.2.5) calculates average forecasting error D of the single model in all test group casesmn, specific formula are as follows:
Wherein | | it is the operation that takes absolute value, DclinIt is the actual dose of voxel, DpredIt is predicted dose, n is single patient
Number of voxel, m are the quantity of test group patient；
(5.2.6) compares the average forecasting error D of each modelmn, the smallest model of average forecasting error is selected, it is corresponding
Hidden layer neuron quantity be selected quantity.
As a preference, in the step (5.3), the training of correlation model the following steps are included:
(5.3.1) filters out the clinical program of same type tumour in case database, the patient's solution for extracting each plan
Cut open feature and dose characteristics；
(5.3.2) inputs the small voxel from different interest regions in the artificial neural network being arranged respectively, instruction
Get the first correlation model of each area-of-interest (ROI)；
(5.3.3) the successively fit solution of each first correlation model of assessment in single plan, filters out first at this
The drill program for being preferably intended to be the ROI refined model is fitted on correlation model, wherein the plan filtered out accounts for general plan
The 70% of quantity；
The refined model of (5.3.4) by the plan chosen in each ROI to the training ROI.
As a preference, assessment models method of fit solution in each plan is in the step (5.3.3), it is first
First calculate the mean absolute error of each case, specific formula are as follows:
Wherein | | it is the operation that takes absolute value, DclinIt is the actual dose value of single voxel, DpredIt is the prediction agent of single voxel
Magnitude, n are the number of voxel of single patient；Illustrate that the fitting effect of the case is better if mean absolute error is smaller, conversely,
Then fitting effect is poorer.
As a preference, using the specific of the 3-dimensional dose distribution of the new patient of correlation model prediction in the step (6)
Step are as follows:
A the anatomical features of new patient) are extracted；
B it) inputs in correlation model and calculates corresponding dose characteristics value；
C new patient) is obtained after the position in CT image is by the dose characteristics value arrangement integration of each voxel according to voxel
Prediction 3-dimensional dose distribution.
The application of a kind of intensity modulated radiation therapy 3-dimensional dose forecast of distribution in the works, using above-mentioned a kind of intensity modulated radiation therapy in the works three
Dosage distribution forecasting method is tieed up, the control of intensity modulated radiation therapy plan quality is carried out.
As a preference, including the following steps:
(A) it for the new patient other than training case, after radiotherapy system (TPS) generates the plan of the patient, extracts
Its anatomical features and dose characteristics, dose characteristics value are the actual dose of current planning；
(B) area-of-interest is successively selected, the anatomical features of the region voxel are inputted into the corresponding correlation model in the region
In, calculate the predicted dose characteristic value in the region；
(C) using dose value as abscissa, percent by volume is ordinate, is drawn respectively according to actual dose and predicted dose
The actual dose volume histogram (DVH) and predicted dose volume histogram of all area-of-interests of new patient；
(D) DVH of each ROI is compared, if there is prediction DVH curve on some or multiple ROI is lower than practical DVH curve,
Then prompting the current planning of the undertaker patient, there are improved spaces.
The working principle of this prediction technique is, from the point of view of adjusting the generation process planned by force, dosage distribution and patient anatomical
There are certain associations between information, and correlation model among these can be calculated by the method for machine learning, new as a result, to suffer from
The dosage distribution of person can predict to come from anatomic information.
Compared with prior art, the present invention have the following advantages that and the utility model has the advantages that
(1) anatomical features of patient can comprehensively be described from volume information, spatial information etc..
(2) three-dimensional dosage distribution is predicted, more dosage informations are presented.
(3) method of machine learning has been borrowed.
Detailed description of the invention
Fig. 1 is intensity modulated radiation therapy 3-dimensional dose distribution forecasting method flow chart in the works.
Fig. 2 is distance-dose relationship figure of bladder voxel, and wherein the distance is minimum range of the voxel to the boundary PTV.
Fig. 3 is distance-dose relationship figure of bladder voxel, and wherein the distance is minimum range of the voxel to rectum boundary.
Fig. 4 is distance-dose relationship figure of bladder voxel, wherein the distance be voxel to left femur boundary in front most narrow spacing
From.
Fig. 5 is distance-dose relationship figure of bladder voxel, and wherein the distance is most narrow spacing of the voxel to right femoral head boundary
From.
Fig. 6 is distance-dose relationship figure of bladder voxel, and wherein the distance is most narrow spacing of the voxel to bulb urethrae boundary
From.
Fig. 7 is the structure chart of neural network.
Fig. 8 is the set interface of neural network.
Fig. 9 is the comparison diagram of practical DVH and prediction DVH in embodiment 2.
Specific embodiment
Method of the invention is described in further detail below with reference to embodiment, but applicable tumor type of the invention
It is not limited to this, without departing from the idea case in the present invention described above, according to ordinary skill knowledge and customary means,
Various replacements and change are made, should all be included within the scope of the invention.
Embodiment 1:
The present embodiment carries out dose prediction to the bladder of patients with prostate cancer, has chosen the VMAT of 14 patients with prostate cancer
It is designed for the training of model.
Firstly, being the voxel of 2.3438 × 2.3438 × 3mm size by the anatomical structure dispersion composition resolution of patient.
Then, the anatomical features and dose characteristics of patient's bladder are extracted using MATLAB, wherein anatomical features include PTV's
Volume, the minimum range of bladder voxel to the boundary PTV, the minimum range of bladder voxel to rectum boundary, bladder voxel to left stock
The minimum range on bone boundary, the minimum range of bladder voxel to right femoral head boundary and bladder voxel are to bulb urethrae boundary
Minimum range, the quantity of model, which is equal to, on single patient jeopardizes organ number plus PTV quantity, as jeopardizes organ number and adds
1.Relationship between the visible each feature of Fig. 2 to Fig. 6 and dosage, in addition to this, dose characteristics are the agent that bladder voxel receives
Magnitude, after this, respectively to each anatomical features value and dose characteristics value linear normalization to [- 1,1] in the range of.
Finally, establish artificial neural network using MATLAB artificial neural network tool box, and train anatomical features and
The correlation model of dose characteristics, detailed process is as follows:
" nntool " order is inputted in MATLAB command window, can pop up the set interface of artificial neural network, at this
In embodiment, three-layer neural network is set；Inputting neuronal quantity is 6；Output neuron quantity is 1；Training function choosing
Use Regularization algorithms；Excitation function selects tanh S type function, and the number of iterations is 500 times；Hidden layer neuron number
The range of amount is selected as 3 to twice of input neuronal quantity, i.e. [3,12], then establishes 10 neural networks, hidden layer
Number is respectively 3 to 12, and 14 cases are randomly divided into training group and test group later, there is 10 cases and 4 cases respectively, will
Training group sequentially inputs 10 networks, trains 10 models, test group is finally sequentially input model, is calculated according to formula 1
Average forecasting error D of each model in all trained cases outmn, specific formula are as follows:
Wherein | | it is the operation that takes absolute value, DclinIt is the actual dose of voxel, DpredIt is predicted dose, n is single patient
Number of voxel, m are the quantity of test group patient.
The smallest model of error is selected, using the hidden layer neuron quantity of this network as the setting of model backward, at this
In embodiment, the quantity of hidden layer neuron is set as 4；Remaining parameter is default.Network structure is as shown in fig. 7, network
Setting is as shown in Figure 8.
The anatomical features of 14 patients and dose characteristics are input to the network set, training obtains first model, is
Fit solution of the model in each case is obtained, the anatomical features of 14 cases are sequentially input in model again, is exported
Dose characteristics value under each case prediction；The mean absolute error of single case is calculated further according to formula 2:
Wherein | | it is the operation that takes absolute value, DclinIt is the actual dose value of single voxel, DpredIt is the prediction agent of single voxel
Magnitude, n are the number of voxel of single patient；Illustrate that the fitting effect of the case is better if mean absolute error is smaller, conversely,
Then fitting effect is poorer.
According to the mean absolute error of each case, training of 10 lesser cases of error as refined model is filtered out
Case, the results are shown in Table 1:
The first model training case verification result table of table 1
Refined model is obtained with the training of garbled case, and calculates the mean absolute error of each trained case, by
The precision of this assessment refined model, the results are shown in Table 2:
2 refined model of table trains case verification result table
Embodiment 2:
The present embodiment by intensity modulated radiation therapy described in above-described embodiment in the works 3-dimensional dose distribution prediction technique based on,
Provide the concrete application method of the prediction technique, specific steps are as follows:
As unit of bladder voxel, 6 anatomical features of new patient's bladder voxel are extracted, volume, voxel including PTV arrive
The minimum range on the boundary PTV, the minimum range of voxel to rectum boundary, voxel to the left femur minimum range on boundary, voxel in front
To right femoral head boundary minimum range and voxel to bulb urethrae boundary minimum range；
Anatomical features are inputted in trained correlation model, the prediction bladder dose characteristics value of new patient is obtained；
(C) using dose value as abscissa, percent by volume is ordinate, is drawn respectively according to actual dose and predicted dose
The actual dose volume histogram (DVH) and predicted dose volume histogram of new patient's bladder, are shown in Fig. 9；
(D) practical bladder DVH curve and prediction bladder DVH curve are compared, if prediction bladder DVH curve is lower than practical DVH
Curve, then it is assumed that there are improved spaces for current planning, and such as the patient 3 in Fig. 9, whole prediction DVH curve is significantly lower than reality
Border DVH curve, can prompt undertaker's current planning that can still advanced optimize.
Although an embodiment of the present invention has been shown and described, it will be understood by those skilled in the art that: not
A variety of change, modification, replacement and modification can be carried out to these embodiments by being detached under the principle of the present invention and objective, of the invention
Range claim and its equivalent unlimited.
Claims (10)
1. a kind of intensity modulated radiation therapy prediction technique that 3-dimensional dose is distributed in the works, is realized by predicting the dosage of each voxel
The prediction of 3-dimensional dose distribution, which comprises the following steps:
(1) it collects effective intensity modulated radiation therapy planning data and forms case database；
(2) according to the resolution sizes of CT image, the target area of patient, each organ that jeopardizes are divided into several small voxels；
(3) extract database in each patient anatomical features, including Gross Target Volume, small voxel to target area boundary minimum range,
Small voxel is to each minimum range for jeopardizing organ boundaries；
(4) dose characteristics of each patient in database, the radiation that the dose characteristics are received by each small voxel are extracted
The complete or collected works of dose value；
(5) artificial neural network is built, the anatomical features and dose characteristics of patient are inputted, goes out to solve by artificial neural network learning
The mapping relations between feature and dose characteristics are cutd open, the correlation model of the two is obtained；
(6) it is distributed using the 3-dimensional dose of the new patient of correlation model prediction.
2. a kind of intensity modulated radiation therapy according to claim 1 prediction technique that 3-dimensional dose is distributed in the works, which is characterized in that
In the step (3), small voxel to target area and each minimum range for jeopardizing organ boundaries be all on three-dimensional space small voxel arrive
Target area or the minimum range for jeopardizing organ boundaries.
3. a kind of intensity modulated radiation therapy according to claim 1 prediction technique that 3-dimensional dose is distributed in the works, which is characterized in that
In the step (3) and (4), place is first normalized before inputting neural network in the anatomical features and dose characteristics of patient
Reason, will be in characteristic value each in anatomical features and dose characteristics value difference Linear Mapping to [- 1,1] range.
4. a kind of intensity modulated radiation therapy according to claim 1 prediction technique that 3-dimensional dose is distributed in the works, which is characterized in that
In the step (5), the foundation of correlation model passes through MATLAB software realization, specifically includes the following steps:
(5.1) artificial neural network tool box is called in MATLAB；
(5.2) neural network is built using MATLAB artificial neural network tool box, which is three-layer neural network；It is defeated
Enter neuronal quantity to be determined by anatomical features quantity；Output neuron quantity is 1；Hidden layer neuron quantity is determined at 3
Into twice of input neuronal quantity, then the specific quantity of hidden layer neuron is determined by model verifying；Training
Function selects Regularization algorithms；Excitation function selects tanh S type function；
(5.3) extracted anatomical features and dose characteristics are inputted into network, trains the correlation model of the two.
5. a kind of intensity modulated radiation therapy according to claim 4 prediction technique that 3-dimensional dose is distributed in the works, which is characterized in that
In the step (5.2), the specific amount of specific steps of hidden layer neuron are determined by model verifying are as follows:
(5.2.1) establishes the different neural network of hidden layer neuron quantity, network number according to hidden layer neuron quantitative range
Amount is determined by neuronal quantity range；
Existing case is divided into training group and test group according to the ratio of 70% and 30% by (5.2.2)；
(5.2.3) sequentially inputs the anatomical features of patient each in training group and dose characteristics in single Neural, training
A model out, in this way, going out the model of respective numbers using all neural metwork trainings；
(5.2.4) sequentially inputs the anatomical features of patient each in test group in single model, obtains each patient single
Predicted dose characteristic value in model so calculates predicted dose characteristic value of each patient in each model；
(5.2.5) calculates average forecasting error D of the single model in all test group casesmn, specific formula are as follows:
Wherein | | it is the operation that takes absolute value, DclinIt is the actual dose of voxel, DpredIt is the predicted dose of voxel, n is single patient
Number of voxel, m is the quantity of test group patient；
(5.2.6) compares the average forecasting error D of each modelmn, the smallest model of average forecasting error is selected, it is corresponding hidden
Hiding layer neuronal quantity is selected quantity.
6. a kind of intensity modulated radiation therapy according to claim 4 prediction technique that 3-dimensional dose is distributed in the works, which is characterized in that
In the step (5.3), the training of correlation model the following steps are included:
(5.3.1) filters out the clinical program of same type tumour in case database, and the patient anatomical for extracting each plan is special
It seeks peace dose characteristics；
(5.3.2) inputs the small voxel from different interest regions in the artificial neural network being arranged respectively, trained
To the first correlation model of each area-of-interest；
(5.3.3) successively assesses fit solution of each first correlation model in single plan, filters out in the first secondary association
The drill program for being preferably intended to be the area-of-interest refined model is fitted on model, wherein the plan filtered out accounts for total
Draw the 70% of quantity；
The refined model of (5.3.4) by the plan chosen in each area-of-interest to the training area-of-interest.
7. a kind of intensity modulated radiation therapy according to claim 6 in the works 3-dimensional dose distribution prediction technique it is characterized in that,
Assessment models method of fit solution in each plan is to calculate being averaged for each case first in the step (5.3.3)
Absolute error, specific formula are as follows:
Wherein | | it is the operation that takes absolute value, DclinIt is the actual dose value of single voxel, DpredIt is the predicted dose of single voxel
Value, n is the number of voxel of single patient；Illustrate that the fitting effect of the case is better if mean absolute error is smaller, conversely,
Fitting effect is poorer.
8. a kind of intensity modulated radiation therapy according to claim 1 prediction technique that 3-dimensional dose is distributed in the works, which is characterized in that
The specific steps being distributed in the step (6) using the 3-dimensional dose of the new patient of correlation model prediction are as follows:
A the anatomical features of new patient) are extracted；
B it) inputs in correlation model and calculates corresponding dose characteristics value；
C the pre- of new patient) is obtained after the position in CT image is by the dose characteristics value arrangement integration of each voxel according to voxel
Survey 3-dimensional dose distribution.
9. a kind of application of intensity modulated radiation therapy 3-dimensional dose forecast of distribution in the works, which is characterized in that using the claims it
A kind of 3-dimensional dose distribution forecasting method in the works of intensity modulated radiation therapy described in one, carries out the control of intensity modulated radiation therapy plan quality.
10. application according to claim 9, which comprises the steps of:
(A) its anatomical features is extracted after radiotherapy system generates the plan of the patient for the new patient other than training case
And dose characteristics, dose characteristics value are the actual dose of current planning；
(B) area-of-interest is successively selected, the anatomical features of the region voxel are inputted in the corresponding correlation model in the region, is counted
Calculate the predicted dose characteristic value in the region；
(C) using dose value as abscissa, percent by volume is ordinate, draws new trouble respectively according to actual dose and predicted dose
The actual dose volume histogram and predicted dose volume histogram of all area-of-interests of person；
(D) dose volume histogram of each area-of-interest is compared, if there is prediction agent on some or multiple semi-cylindrical hills
It measures volume histogram and is lower than actual dose volume histogram, then prompting the current planning of the undertaker patient, there are improved skies
Between.
Applications Claiming Priority (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2017107605526 | 2017-08-30 | ||
| CN201710760552 | 2017-08-30 |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN107441637A CN107441637A (en) | 2017-12-08 |
| CN107441637B true CN107441637B (en) | 2019-06-07 |
Family
ID=60497219
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710851175.7A Active CN107441637B (en) | 2017-08-30 | 2017-09-20 | Intensity modulated radiation therapy 3-dimensional dose is distributed in the works prediction technique and its application |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107441637B (en) |
Families Citing this family (24)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109979564A (en) * | 2017-12-28 | 2019-07-05 | 北京连心医疗科技有限公司 | A kind of intelligence radiotherapy planning method, equipment and storage medium |
| CN108364678B (en) * | 2018-02-10 | 2022-02-01 | 福建省肿瘤医院（福建省肿瘤研究所、福建省癌症防治中心） | Nasopharyngeal carcinoma radiotherapy plan automatic design method based on machine learning |
| CN108717866B (en) * | 2018-04-03 | 2022-10-11 | 中国医学科学院肿瘤医院 | Method, device, equipment and storage medium for predicting radiotherapy plan dose distribution |
| US11557390B2 (en) * | 2018-04-30 | 2023-01-17 | Elekta, Inc. | Radiotherapy treatment plan modeling using generative adversarial networks |
| CN108648811A (en) * | 2018-05-11 | 2018-10-12 | 南方医科大学 | A kind of automatic radiotherapy planning quality evaluating method based on the prediction of indication item |
| CN108899093A (en) * | 2018-05-11 | 2018-11-27 | 南方医科大学 | Intensity modulated radiation therapy intended dose indication item prediction technique based on intelligence learning |
| CN109166613A (en) * | 2018-08-20 | 2019-01-08 | 北京东方瑞云科技有限公司 | Radiotherapy treatment planning assessment system and method based on machine learning |
| CN111035860B (en) * | 2018-10-11 | 2022-02-18 | 海创时代(深圳)医疗科技有限公司 | Automatic optimization device of planning system |
| CN109754860B (en) * | 2018-12-21 | 2020-11-10 | 苏州雷泰医疗科技有限公司 | Method for automatically predicting planning difficulty level based on DVH prediction model |
| CN109859817B (en) | 2019-04-01 | 2022-12-09 | 安徽大学 | Method and system for evaluating risk organ dosimetry in esophageal radiotherapy plan |
| CN110085298B (en) * | 2019-04-26 | 2022-02-01 | 南方医科大学 | Intensity modulated radiotherapy plan three-dimensional dose distribution prediction method based on deep network learning |
| CN110141801B (en) * | 2019-05-17 | 2021-04-27 | 天津大学 | Intelligent dose planning method for close-range intra-particle radiotherapy operation |
| CN110124214B (en) * | 2019-05-24 | 2020-09-04 | 南方医科大学 | Predictive dose distribution guided intensity modulated radiotherapy plan optimization systems, media and devices |
| CN110197709B (en) * | 2019-05-29 | 2023-06-20 | 广州瑞多思医疗科技有限公司 | Three-dimensional dose prediction method based on deep learning and priori planning |
| FR3095332B1 (en) * | 2019-06-27 | 2023-07-21 | Quantum Surgical | Tissue ablation planning method based on deep learning |
| CN110327554B (en) * | 2019-07-08 | 2020-11-10 | 南方医科大学 | Intensity modulated radiotherapy plan optimization method based on predicted dose distribution guidance and application |
| CN110433405A (en) * | 2019-07-09 | 2019-11-12 | 苏州雷泰智能科技有限公司 | A kind of TPS optimization method and device of predicted dose guidance |
| CN110354406A (en) * | 2019-07-30 | 2019-10-22 | 安徽大学 | A kind of the 3-dimensional dose prediction technique and system of radiotherapy |
| CN110464353A (en) * | 2019-08-21 | 2019-11-19 | 南方医科大学 | A kind of pseudo- CT synthetic method and application based on depth convolutional neural networks |
| US11077320B1 (en) | 2020-02-07 | 2021-08-03 | Elekta, Inc. | Adversarial prediction of radiotherapy treatment plans |
| CN111951245B (en) * | 2020-08-11 | 2021-04-06 | 山东省肿瘤防治研究院（山东省肿瘤医院） | Method for determining radiation therapy dosage according to characteristic parameters of tumor molecular image |
| CN113101548B (en) * | 2021-04-20 | 2023-04-28 | 中山大学肿瘤防治中心 | Photon intensity modulated radiotherapy control method for reducing skin dose |
| US20240001138A1 (en) * | 2022-06-29 | 2024-01-04 | Siemens Healthineers International Ag | Detecting anomalous dose volume histogram information |
| CN115300811B (en) * | 2022-08-08 | 2024-01-05 | 中山大学肿瘤防治中心（中山大学附属肿瘤医院、中山大学肿瘤研究所） | Dose distribution determining method and device based on machine learning |
Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104117151A (en) * | 2014-08-12 | 2014-10-29 | 章桦 | Optimization method of online self-adaption radiotherapy plan |
Family Cites Families (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP1909904B1 (en) * | 2005-07-25 | 2013-09-04 | Karl Otto | Methods and apparatus for the planning of radiation treatments |
| EP2585854B1 (en) * | 2010-06-22 | 2020-03-18 | Varian Medical Systems International AG | System and method for estimating and manipulating estimated radiation dose |
| CN104645500B (en) * | 2015-02-12 | 2018-08-03 | 上海联影医疗科技有限公司 | A kind of intensity modulated radiation therapy optimization system |
-
2017
- 2017-09-20 CN CN201710851175.7A patent/CN107441637B/en active Active
Patent Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104117151A (en) * | 2014-08-12 | 2014-10-29 | 章桦 | Optimization method of online self-adaption radiotherapy plan |
Also Published As
| Publication number | Publication date |
|---|---|
| CN107441637A (en) | 2017-12-08 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107441637B (en) | Intensity modulated radiation therapy 3-dimensional dose is distributed in the works prediction technique and its application | |
| US10328282B2 (en) | System and method for novel chance-constrained optimization in intensity-modulated proton therapy planning to account for range and patient setup uncertainties | |
| US10265543B2 (en) | Beam segment-level dose computation and temporal motion tracking for adaptive treatment planning | |
| US9744379B2 (en) | Studying dosimetric impact of motion to generate adaptive patient-specific margins in EBRT planning | |
| CN114681813B (en) | Automatic radiation therapy planning system, automatic radiation therapy planning method, and storage medium | |
| US11738211B2 (en) | Dose-distribution estimation in proton therapy | |
| EP1708788A2 (en) | Systems and methods for global optimization of treatment planning for external beam radiation therapy | |
| CN110327554B (en) | Intensity modulated radiotherapy plan optimization method based on predicted dose distribution guidance and application | |
| WO2016144915A1 (en) | Automatic determination of radiation beam configurations for patient-specific radiation therapy planning | |
| CN110141801A (en) | Towards close-range particle Inner irradiation operation smart dosage planing method | |
| US11724126B2 (en) | Methods and systems for quality-aware continuous learning for radiotherapy treatment planning | |
| Biston et al. | Time of PTV is ending, robust optimization comes next | |
| Wu et al. | An evaluation of planning techniques for stereotactic body radiation therapy in lung tumors | |
| Yousefi et al. | A novel mathematical model to generate semi‐automated optimal IMRT treatment plan based on predicted 3D dose distribution and prescribed dose | |
| Leal et al. | MLC leaf width impact on the clinical dose distribution: a Monte Carlo approach | |
| Yang et al. | Provision of organ at risk contouring guidance in UK radiotherapy clinical trials | |
| Peng et al. | Toward real-time automatic treatment planning (RTTP) with a one-step 3D fluence map prediction method and (nonorthogonal) convolution technique | |
| CN113178242A (en) | Automatic plan optimization system based on coupled generation countermeasure network | |
| Robinson et al. | Can the use of knowledge-based planning systems improve stereotactic radiotherapy planning? A systematic review | |
| Nasser et al. | A head and neck treatment planning strategy for a CBCT‐guided ring‐gantry online adaptive radiotherapy system | |
| Christiansen | A framework for the robust delivery of respiratory motion adaptive arc radiotherapy | |
| Voet | Automation of contouring and planning in radiotherapy | |
| Sprouts | Advancing the Radiation Oncology Clinic with Motion Management and Automatic Treatment Planning | |
| Omotayo et al. | Feasibility study for marker‐based VMAT plan optimization toward tumor tracking | |
| Wang | Automated Generation of Radiotherapy Treatment Plans Using Machine Learning Methods |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |