CN109166613A - Radiotherapy treatment planning assessment system and method based on machine learning - Google Patents
Radiotherapy treatment planning assessment system and method based on machine learning Download PDFInfo
- Publication number
- CN109166613A CN109166613A CN201810948783.4A CN201810948783A CN109166613A CN 109166613 A CN109166613 A CN 109166613A CN 201810948783 A CN201810948783 A CN 201810948783A CN 109166613 A CN109166613 A CN 109166613A
- Authority
- CN
- China
- Prior art keywords
- voxel
- treatment planning
- radiotherapy treatment
- dose
- machine learning
- 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.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G16H20/00—ICT specially adapted for therapies or health-improving plans, e.g. for handling prescriptions, for steering therapy or for monitoring patient compliance
- G16H20/40—ICT specially adapted for therapies or health-improving plans, e.g. for handling prescriptions, for steering therapy or for monitoring patient compliance relating to mechanical, radiation or invasive therapies, e.g. surgery, laser therapy, dialysis or acupuncture
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G16H70/00—ICT specially adapted for the handling or processing of medical references
Landscapes
- Health & Medical Sciences (AREA)
- Engineering & Computer Science (AREA)
- Epidemiology (AREA)
- General Health & Medical Sciences (AREA)
- Medical Informatics (AREA)
- Primary Health Care (AREA)
- Public Health (AREA)
- Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
- Surgery (AREA)
- Urology & Nephrology (AREA)
- Radiation-Therapy Devices (AREA)
Abstract
The present invention relates to radiotherapy treatment planning evaluation areas, the radiotherapy treatment planning appraisal procedure based on machine learning that it discloses a kind of solves the problems, such as occur artificial subjective bias there are standard disunity and easily using manual evaluation radiotherapy treatment planning in traditional technology.This method can be summarized as: carry out the pretreatment works such as voxel selection, voxel feature extraction and voxel data mark to the DICOM data of the radiotherapy history plan of one group of high quality first; then using voxel data one machine learning model of training; this model is that each voxel in the works to be assessed exports predicted dose value; and two-dimensional DVH prediction curve and three-dimensional voxel dose prediction distribution figure further are generated for each crisis organ, last physics teacher carries out actual assessment to radiotherapy planning to be assessed with reference to this group of curve and distribution map.In addition, being suitable for carrying out objective, accurate evaluation to radiotherapy planning the invention also discloses corresponding assessment system.
Description
Technical field
The present invention relates to radiotherapy treatment planning evaluation areas, and in particular to a kind of radiotherapy treatment planning based on machine learning
Assessment system and method.
Background technique
Radiotherapy is a kind of local therapeutic approaches using radiation cure tumour, is the oncotherapy of current mainstream
One of mode.Radioactive ray include that α, β, gamma-rays and all kinds of x-ray therapy apparatuses that radioactive isotope generates or accelerator generate
X-ray, electric wire, proton beam and other particles beams etc..Currently, about 70% cancer patient is during treating cancer
It needs using radiotherapy.
Radiotherapy treatment planning, be control accelerator irradiation patient's knub position specific embodiment, by physics teacher according to
The prescribed dose target that radiotherapy doctor specifies is made using treatment planning systems (Treatment Planning System, TPS)
Fixed, target is to control organs in adult (i.e. as far as possible while guaranteeing that tumour (i.e. target area) obtains enough Radiotherapy dosimetries
Jeopardize organ) exposure dose be no more than its tolerance dose, to protect the function of organs in adult and the life matter of patient
Amount.
Radiotherapy treatment planning assessment, refer to before actual implementation radiotherapy to physics teacher work out radiotherapy treatment planning into
Row quality evaluation, to decide whether using this radiotherapy treatment planning.The general standard of Program Assessment is that target dose meets
Prescribed dose, and the dosage for jeopardizing organ is low as much as possible.But the target location and differences in shape due to each patient compared with
Greatly, the location and shape for jeopardizing organ also have certain difference, and the degree that jeopardizing the dosage of organ can reach is also to have larger difference
Other, therefore, when practical Program Assessment, the height judgement for jeopardizing the dose value of organ is exactly a very doubt job.Mesh
Before, rely primarily on veteran physics teacher rule of thumb with comprehensive judgement is carried out the case where specific patient, it is more time-consuming and laborious,
And the standard of each physics teacher also disunity, there is also the deviation of artificial subjectivity sometimes.
Summary of the invention
The technical problems to be solved by the present invention are: proposing a kind of radiotherapy treatment planning assessment system based on machine learning
And method, it solves that manual evaluation radiotherapy treatment planning is used to occur artificial subjectivity there are standard disunity and easily partially in traditional technology
The problem of difference.
The technical proposal adopted by the invention to solve the above technical problems is that:
Radiotherapy treatment planning assessment system based on machine learning, comprising:
Sample preprocessing module, the DICOM (medical digital images for the history radiotherapy treatment planning from one group of high quality
And communication) carry out voxel (volume element: being numerical data in the minimum unit in three-dimensional space segmentation) in data and choose, voxel
Feature extraction and voxel data mark, export voxel sample；
Assessment models training module, for one machine learning of voxel sample training using the output of sample preprocessing module
Model is as radiotherapy treatment planning assessment models；
Prediction module, for using trained radiotherapy treatment planning assessment models to radiotherapy treatment planning number to be assessed
Voxel in carries out dose prediction, and further jeopardizes organ for each and generate two dimension DVH (dose volume histogram) in advance
Survey curve and three-dimensional voxel dose prediction distribution map；
Actual assessment module, for dividing for physics teacher with reference to the two dimension DVH prediction curve and three-dimensional voxel dose prediction
Butut carries out actual assessment to radiotherapy treatment planning to be assessed.
As advanced optimizing, the sample preprocessing module includes:
History radiotherapy treatment planning determining module, for determining the history radiotherapy treatment planning of one group of high quality；
Voxel choose module, for from the DICOM data of the history radiotherapy treatment planning of one group of determining high quality into
Row voxel is chosen: choosing the number of voxels that the voxel around target area within the scope of certain distance is used as training machine learning model
According to voxel data includes the voxel for jeopardizing organ；
Voxel characteristic extracting module extracts characteristic, characteristic packet for the voxel to selection from DICOM data
Include but be not limited to: PTV volume, the shortest distance of the target area voxel distance PTV and voxel distance jeopardize the organ shortest distance；
Voxel data labeling module, for marking a dose value according to voxel of the DICOM data to each selection；
Voxel sample output module exports the voxel sample for the voxel after marking as voxel sample.
As advanced optimizing, the machine learning model of the assessment models training module training is based on supervised learning side
A kind of machine learning model of formula；
The machine learning model can be used following functional relation and briefly express:
Y=h (X)
Wherein, h function is the functional transformation relationship of machine learning model expression, and X is the characteristic of the voxel of input, Y
It is the dose prediction value of the voxel of output.
As advanced optimizing, the actual assessment that the actual assessment module carries out is difference assessment, i.e., puts to be assessed
The dose value penetrated in treatment plan and be compared by the dose value of assessment models prediction output, if it is to be assessed in the works
Jeopardize the dose value of organ below or near to the dose value of assessment models prediction output, then shows that treatment plan quality is higher；Such as
The fruit dose value for jeopardizing organ in the works to be assessed is higher than the dose value of assessment models prediction output, then shows treatment plan matter
Amount is not high, and physics teacher is needed to advanced optimize radiotherapy treatment planning.
As advanced optimizing, the prediction module is specifically used for:
Using the characteristic data value of the voxel of plan to be assessed as input value, input in radiotherapy treatment planning assessment models,
By model prediction, the dose prediction value of voxel is exported, is further formed two-dimensional DVH prediction curve and three-dimensional after processing
Voxel dose prediction distribution figure.
In addition, being based on above-mentioned assessment system, the present invention also provides a kind of radiotherapy treatment plannings based on machine learning to comment
Estimate method comprising following steps:
A. voxel selection is carried out from the DICOM data of the history radiotherapy treatment planning of one group of high quality, voxel feature mentions
It takes and is marked with voxel data, obtain voxel sample；
B. using one machine learning model of voxel sample training as radiotherapy treatment planning assessment models；
C. using trained radiotherapy treatment planning assessment models to the voxel in radiotherapy treatment planning data to be assessed
Dose prediction is carried out, and further jeopardizes organ for each and generates two dimension DVH prediction curve and three-dimensional voxel dose prediction point
Butut；
D. physics teacher is with reference to the two dimension DVH prediction curve and three-dimensional voxel dose prediction distribution map to radiation to be assessed
Treatment plan carries out actual assessment.
As advanced optimizing, step a is specifically included:
By experienced physics teacher according to corresponding conditions comprehensive descision, one group is chosen from history radiotherapy treatment planning；
Wherein, corresponding conditions include: that target dose reaches prescribed dose, and target dose will uniformly, and it is low as much as possible to jeopardize organ dose,
The dosage behind target area, which falls speed, out to jeopardize the hot spot that cannot have dosage excessively high in organ fastly；
Then the DICOM data of selected history radiotherapy treatment planning are proceeded as follows:
A1. voxel is chosen: choosing the voxel around target area within the scope of certain distance and uses as training machine learning model
Voxel data, voxel data includes the voxel for jeopardizing organ；
A2. voxel feature extraction: extracting characteristic to the voxel of selection from DICOM data, characteristic include but
Be not limited to: PTV volume, the shortest distance of the target area voxel distance PTV and voxel distance jeopardize the organ shortest distance；
A3. voxel data marks: marking a dose value according to voxel of the DICOM data to each selection.
As advanced optimizing, in step b, the machine learning model is a kind of engineering based on supervised learning mode
Practise model；
The machine learning model can be used following functional relation and briefly express:
Y=h (X)
Wherein, h function is the functional transformation relationship of machine learning model expression, and X is the characteristic of the voxel of input, Y
It is the dose prediction value of the voxel of output.
As advanced optimizing, in step c, the input value of the radiotherapy treatment planning assessment models is plan to be assessed
The characteristic data value of voxel, output valve are the dose prediction value of voxel, are further formed two-dimensional DVH prediction curve after processing
With three-dimensional voxel dose prediction distribution figure.
As advanced optimizing, in step d, the actual assessment is difference assessment, i.e., by radiotherapy treatment planning to be assessed
In dose value and be compared by the dose value of assessment models prediction output, if it is to be assessed in the works jeopardize organ
Dose value then shows that treatment plan quality is higher below or near to the dose value of assessment models prediction output；If estimation to be evaluated
The dose value for jeopardizing organ in drawing is higher than the dose value of assessment models prediction output, then shows that treatment plan is of low quality, needs
Physics teacher is wanted to advanced optimize radiotherapy treatment planning.
The beneficial effects of the present invention are:
1) can obtain more objective and accurate radiotherapy treatment planning Evaluated effect: the method for manual evaluation is vulnerable to subjectivity
It influences, it is possible that deviation when assessment；And the present invention uses history planning data middle school of the machine learning model from high quality
The dosage regularity of distribution is practised, more objective and accurate dose prediction can be obtained as a result, obtaining more objectively and accurately assessing in turn
Plan quality greatly reduces subjectivity and a possibility that deviation occurs.
2) can save the Program Assessment workload of senior physics Shi Chang: manual evaluation plan is one cumbersome time-consuming
Work, it is higher to the skill requirement of physics teacher, the plan that new physics teacher does generally require senior physics teacher devote considerable time and
Energy is assessed；And the present invention does not depend on the experience of senior physics teacher, can accomplish automation assessment, to save senior
The Program Assessment workload of physics Shi Chang.
3) scheme has versatility: the making step and last minute planning effect of the treatment planning systems of different manufacturers all slightly have
Difference may will affect final Evaluated effect；And the present invention relies only on the DICOM data of history plan, is not rely on spy
Fixed treatment planning systems, this independence make it with more versatility.
Detailed description of the invention
Fig. 1 is the radiotherapy treatment planning assessment system structural block diagram based on machine learning in the present invention；
Fig. 2 is the radiotherapy treatment planning appraisal procedure flow chart based on machine learning in the present invention.
Specific embodiment
The present invention is directed to propose a kind of radiotherapy treatment planning assessment system and method based on machine learning, solves traditional skill
There are problems that standard disunity using manual evaluation radiotherapy treatment planning in art and artificial subjective bias easily occur.
Machine learning is the visitor for learning the objective law in certain field from historical data, and being arrived using this automatic study
See a kind of method that rule predicts new data.Common machine learning method is divided into supervised learning mode and unsupervised
Mode of learning, wherein supervised learning mode is at present using a kind of more mode.The general step of supervised learning mode is,
Firstly, extracting one group of characteristic to historical data and being labeled to historical data, then, characteristic and mark number are used
It will be to be predicted after the completion of training according to one machine learning model (for example, neural network, linear regression, random forest etc.) of training
Data characteristic input model, model can export a prediction result value.This prediction result value embodies history number
The rule for including in, accuracy with higher.
The present invention is exactly to be learnt from the history planning data of high quality using the machine learning model of supervised learning mode
The dosage regularity of distribution is conducive to pair to carry out dose prediction to the voxel in radiotherapy treatment planning data to be assessed automatically
Radiotherapy treatment planning more efficiently, accurately, objectively assess.
As shown in Figure 1, the radiotherapy treatment planning assessment system based on machine learning in the present invention includes sample preprocessing
Four module, assessment models training module, prediction module and actual assessment module parts；The function of modules is as follows:
Sample preprocessing module, specifically includes: history radiotherapy treatment planning determining module, voxel choose module, voxel spy
Levy extraction module, voxel data labeling module and voxel sample output module；Wherein, history radiotherapy treatment planning determining module,
For determining the history radiotherapy treatment planning of one group of high quality；Voxel chooses module, for going through from one group of determining high quality
Voxel selection is carried out in the DICOM data of history radiotherapy treatment planning: choose around the target area PTV in certain distance (such as: 30mm)
The voxel data that is used as training machine learning model of voxel, voxel data includes the voxel for jeopardizing organ；Voxel feature
Extraction module extracts characteristic for the voxel to selection from DICOM data, and characteristic includes but is not limited to: PTV body
Product, the shortest distance of the target area voxel distance PTV and voxel distance jeopardize the organ shortest distance；Voxel data labeling module is used
In voxel one dose value of mark according to DICOM data to each selection；Voxel sample output module, for it will mark after
Voxel exports the voxel sample as voxel sample.
Assessment models training module, for one machine learning of voxel sample training using the output of sample preprocessing module
Model is as radiotherapy treatment planning assessment models；
Prediction module, for using trained radiotherapy treatment planning assessment models to radiotherapy treatment planning number to be assessed
Voxel in carries out dose prediction, and further jeopardizes organ for each and generate two dimension DVH prediction curve and three-dimensional voxel
Dose prediction distribution map；
Actual assessment module, for dividing for physics teacher with reference to the two dimension DVH prediction curve and three-dimensional voxel dose prediction
Butut carries out actual assessment to radiotherapy treatment planning to be assessed.
Based on above system, the radiotherapy treatment planning appraisal procedure based on machine learning that the present invention realizes as shown in Fig. 2,
It includes following implementation steps:
1, voxel selection is carried out from the DICOM data of the history radiotherapy treatment planning of one group of high quality, voxel feature mentions
It takes and is marked with voxel data, obtain voxel sample；
In this step, it is first determined the history radiotherapy treatment planning of one group of high quality, in specific implementation, can by have through
The physics teacher tested chooses 100 according to corresponding conditions comprehensive descision from history radiotherapy treatment planning；Wherein, corresponding conditions packet
Include: target dose reaches prescribed dose, and target dose uniformly will jeopardize that organ dose is low as much as possible, out the dosage behind target area
It is fast to fall speed, jeopardizes the hot spot etc. that there cannot be dosage excessively high in organ.
Then the DICOM data of selected history radiotherapy treatment planning are proceeded as follows:
Voxel is chosen: the voxel data that the voxel of 30mm around the target area PTV is used as training machine learning model is chosen,
Voxel data can only include the voxel for jeopardizing organ, also may include the voxel for jeopardizing organ and other tissues；
Voxel feature extraction: characteristic is extracted from DICOM data to the voxel of selection, characteristic includes PTV body
Product, the shortest distance, the voxel distance of the target area voxel distance PTV jeopardize organ shortest distance etc.；
Voxel data mark: a dose value is marked according to each voxel of the DICOM data to selection.
2, using one machine learning model of voxel sample training as radiotherapy treatment planning assessment models；
In this step, using one machine learning model of voxel sample training is obtained in step 1, machine learning model is used
Neural network model；The effect of entire neural network model can be briefly expressed with following functional relation:
Y=h (X)
Wherein, h function is the functional transformation relationship of neural network expression, and X is the characteristic of the voxel of input, and Y is defeated
The dose prediction value of voxel out.The cost function used when neural metwork training for intersect entropy function, neural network parameter
Optimization method is gradient descent algorithm, and the algorithm for solving gradient is back-propagation algorithm.
Certainly, machine learning model here can also use other models based on supervised learning mode, such as: linear to return
Return model, Random Forest model etc..
3, using trained radiotherapy treatment planning assessment models to the voxel in radiotherapy treatment planning data to be assessed
Dose prediction is carried out, and further jeopardizes organ for each and generates two dimension DVH prediction curve and three-dimensional voxel dose prediction point
Butut；
In this step, using the characteristic data value of the voxel of plan to be assessed as input, trained radiotherapy is utilized
Program Assessment model is predicted, the dose prediction value of voxel is exported, due to have been obtained for said three-dimensional body (human body) each
The predicted dose numerical value of voxel, is shown by 3-D image, can obtain three-dimensional voxel dose prediction distribution map；In addition, according to
The dose value of each voxel jeopardizes the information of organ in conjunction with belonging to voxel, and it is bent can also directly to calculate two-dimentional DVH prediction
Line, specific calculation belong to the common knowledge of field of radiation therapy, and the present invention just no longer repeats it.
4, physics teacher is with reference to the two dimension DVH prediction curve and three-dimensional voxel dose prediction distribution map to radiation to be assessed
Treatment plan carries out actual assessment.
In this step, the actual assessment is difference assessment, i.e., by dose value in radiotherapy treatment planning to be assessed and logical
The dose value for crossing assessment models prediction output be compared (when practical operation, two dimension that physics teacher will be exported by assessment models
The two dimension generated when DVH prediction curve and three-dimensional voxel dose prediction distribution map are with by using TPS system formulation radiotherapy planning
DVH prediction curve and three-dimensional voxel dose prediction distribution map are compared), if the dosage for jeopardizing organ in the works to be assessed
Value then shows that treatment plan quality is higher, without adjusting treatment plan below or near to the dose value of assessment models prediction output；
If the dose value for jeopardizing organ in the works to be assessed is higher than the dose value of assessment models prediction output, show treatment plan
It is of low quality, need physics teacher to advanced optimize radiotherapy treatment planning.
Claims (10)
1. the radiotherapy treatment planning assessment system based on machine learning characterized by comprising
Sample preprocessing module, for carrying out voxel choosing from the DICOM data of the history radiotherapy treatment planning of one group of high quality
It takes, voxel feature extraction and voxel data mark, exports voxel sample；
Assessment models training module, for one machine learning model of voxel sample training using the output of sample preprocessing module
As radiotherapy treatment planning assessment models；
Prediction module, for using trained radiotherapy treatment planning assessment models in radiotherapy treatment planning data to be assessed
Voxel carry out dose prediction, and further jeopardize organ for each and generate two dimension DVH prediction curve and three-dimensional voxel dosage
Prediction distribution figure；
Actual assessment module, for referring to the two dimension DVH prediction curve and three-dimensional voxel dose prediction distribution map for physics teacher
Actual assessment is carried out to radiotherapy treatment planning to be assessed.
2. the radiotherapy treatment planning assessment system based on machine learning as described in claim 1, which is characterized in that
The sample preprocessing module includes:
History radiotherapy treatment planning determining module, for determining the history radiotherapy treatment planning of one group of high quality；
Voxel chooses module, for carrying out body from the DICOM data of the history radiotherapy treatment planning of one group of determining high quality
Element is chosen: choosing the voxel data that the voxel around target area within the scope of certain distance is used as training machine learning model, body
Prime number is according to the voxel including jeopardizing organ；
Voxel characteristic extracting module extracts characteristic for the voxel to selection from DICOM data, and characteristic includes
PTV volume, the shortest distance of the target area voxel distance PTV and voxel distance jeopardize the organ shortest distance；
Voxel data labeling module, for marking a dose value according to voxel of the DICOM data to each selection；
Voxel sample output module exports the voxel sample for the voxel after marking as voxel sample.
3. the radiotherapy treatment planning assessment system based on machine learning as described in claim 1, which is characterized in that
The machine learning model of the assessment models training module training is a kind of machine learning mould based on supervised learning mode
Type,
The machine learning model can be used following functional relation and briefly express:
Y=h (X)
Wherein, h function is the functional transformation relationship of machine learning model expression, and X is the characteristic of the voxel of input, and Y is defeated
The dose prediction value of voxel out.
4. the radiotherapy treatment planning assessment system based on machine learning as described in claim 1, which is characterized in that
The actual assessment that the actual assessment module carries out is difference assessment, i.e., by the dose value in radiotherapy treatment planning to be assessed
It is compared with the dose value by assessment models prediction output, if the dose value for jeopardizing organ in the works to be assessed is lower than
Or the dose value close to assessment models prediction output, then show that treatment plan quality is higher；If jeopardizing in the works to be assessed
The dose value of organ is higher than the dose value of assessment models prediction output, then shows that treatment plan is of low quality, need physics teacher couple
Radiotherapy treatment planning advanced optimizes.
5. the radiotherapy treatment planning assessment system based on machine learning as described in claim 1-4 any one, feature exist
In the prediction module is specifically used for:
Using the characteristic data value of the voxel of plan to be assessed as input value, inputs in radiotherapy treatment planning assessment models, pass through
Model prediction exports the dose prediction value of voxel, is further formed two-dimensional DVH prediction curve and three-dimensional voxel after processing
Dose prediction distribution map.
6. the radiotherapy treatment planning appraisal procedure based on machine learning, which comprises the following steps:
A. from the DICOM data of the history radiotherapy treatment planning of one group of high quality carry out voxel selection, voxel feature extraction and
Voxel data mark, obtains voxel sample；
B. using one machine learning model of voxel sample training as radiotherapy treatment planning assessment models；
C. the voxel in radiotherapy treatment planning data to be assessed is carried out using trained radiotherapy treatment planning assessment models
Dose prediction, and further jeopardize organ for each and generate two dimension DVH prediction curve and three-dimensional voxel dose prediction distribution map；
D. physics teacher is with reference to the two dimension DVH prediction curve and three-dimensional voxel dose prediction distribution map to radiotherapy to be assessed
Plan carries out actual assessment.
7. the radiotherapy treatment planning appraisal procedure based on machine learning as claimed in claim 6, which is characterized in that
Step a is specifically included:
By experienced physics teacher according to corresponding conditions comprehensive descision, one group is chosen from history radiotherapy treatment planning；Wherein,
Corresponding conditions include: that target dose reaches prescribed dose, and target dose will uniformly, and it is low as much as possible to jeopardize organ dose, out target
Dosage behind area, which falls speed, to jeopardize the hot spot that cannot have dosage excessively high in organ fastly；
Then the DICOM data of selected history radiotherapy treatment planning are proceeded as follows:
A1. voxel is chosen: choosing the body that the voxel around target area within the scope of certain distance is used as training machine learning model
Prime number evidence, voxel data include the voxel for jeopardizing organ；
A2. voxel feature extraction: characteristic is extracted from DICOM data to the voxel of selection, characteristic includes PTV body
Product, the shortest distance of the target area voxel distance PTV and voxel distance jeopardize the organ shortest distance；
A3. voxel data marks: marking a dose value according to voxel of the DICOM data to each selection.
8. the radiotherapy treatment planning appraisal procedure based on machine learning as claimed in claim 6, which is characterized in that
In step b, the machine learning model is a kind of machine learning model based on supervised learning mode；
The machine learning model can be used following functional relation and briefly express:
Y=h (X)
Wherein, h function is the functional transformation relationship of machine learning model expression, and X is the characteristic of the voxel of input, and Y is defeated
The dose prediction value of voxel out.
9. the radiotherapy treatment planning appraisal procedure based on machine learning as claimed in claim 6, which is characterized in that
In step c, the input value of the radiotherapy treatment planning assessment models is the characteristic data value of the voxel of plan to be assessed, defeated
Value is the dose prediction value of voxel out, is further formed two-dimensional DVH prediction curve after processing and three-dimensional voxel dose is pre-
Survey distribution map.
10. the radiotherapy treatment planning appraisal procedure based on machine learning as described in claim 6-9 any one, feature exist
In,
In step d, the actual assessment is difference assessment, i.e., by the dose value in radiotherapy treatment planning to be assessed and passes through assessment
The dose value of model prediction output is compared, if the dose value for jeopardizing organ in the works to be assessed is below or near to assessment
The dose value of model prediction output, then show that treatment plan quality is higher；If the dosage for jeopardizing organ in the works to be assessed
Value is higher than the dose value of assessment models prediction output, then shows that treatment plan is of low quality, need physics teacher to radiotherapy meter
It draws and advanced optimizes.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810948783.4A CN109166613A (en) | 2018-08-20 | 2018-08-20 | Radiotherapy treatment planning assessment system and method based on machine learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810948783.4A CN109166613A (en) | 2018-08-20 | 2018-08-20 | Radiotherapy treatment planning assessment system and method based on machine learning |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109166613A true CN109166613A (en) | 2019-01-08 |
Family
ID=64896266
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810948783.4A Pending CN109166613A (en) | 2018-08-20 | 2018-08-20 | Radiotherapy treatment planning assessment system and method based on machine learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109166613A (en) |
Cited By (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109859817A (en) * | 2019-04-01 | 2019-06-07 | 安徽大学 | Oesophagus radiotherapy treatment planning risk organ dose learns appraisal procedure, assessment system |
| CN110197709A (en) * | 2019-05-29 | 2019-09-03 | 广州瑞多思医疗科技有限公司 | A kind of 3-dimensional dose prediction technique based on deep learning Yu priori plan |
| CN110232964A (en) * | 2019-06-04 | 2019-09-13 | 苏州雷泰智能科技有限公司 | Plan implementation method and device based on predicted dose guidance and Gaussian process optimization |
| CN110841205A (en) * | 2019-10-21 | 2020-02-28 | 温州医科大学附属第一医院 | Accurate dose verification method, device and equipment for tumor patient |
| CN111540437A (en) * | 2020-04-23 | 2020-08-14 | 北京大学第三医院（北京大学第三临床医学院） | Dose verification method and system based on artificial intelligence |
| WO2021090063A1 (en) * | 2019-11-08 | 2021-05-14 | Wang Qunxi | Cloud-based dose verification |
| CN114007538A (en) * | 2019-06-27 | 2022-02-01 | 康坦手术股份有限公司 | Deep learning-based method for planning tissue ablation |
| JP2022530413A (en) * | 2019-09-26 | 2022-06-29 | ベイジン・センスタイム・テクノロジー・デベロップメント・カンパニー・リミテッド | Image processing methods and equipment, electronic devices, storage media and computer programs |
| CN115223683A (en) * | 2022-08-19 | 2022-10-21 | 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） | Radiotherapy plan evaluation method, radiotherapy plan evaluation device, computer equipment and medium |
| CN115862812A (en) * | 2023-02-28 | 2023-03-28 | 福建自贸试验区厦门片区Manteia数据科技有限公司 | Radiotherapy plan processing device, storage medium and electronic device |
| CN116569190A (en) * | 2020-12-09 | 2023-08-08 | 光线搜索实验室公司 | Method and computer program product for dose map prediction for radiotherapy treatment planning |
Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104225806A (en) * | 2014-08-28 | 2014-12-24 | 高献书 | Radiotherapy plan design method and system based on biologically effective dose |
| CN104815392A (en) * | 2015-04-10 | 2015-08-05 | 石峰 | Interactive radiotherapy plan system optimization method and system thereof |
| WO2017063691A1 (en) * | 2015-10-14 | 2017-04-20 | Brainlab Ag | Normal tissue sparing in radiation therapy treatment planning using prioritization for optimizing tissue indices |
| CN106902478A (en) * | 2017-02-22 | 2017-06-30 | 天津大学 | A kind of systematization radiotherapy biological effect appraisal procedure |
| CN107072624A (en) * | 2014-06-18 | 2017-08-18 | Elekta股份有限公司 | system and method for automatic treatment plan |
| CN107441637A (en) * | 2017-08-30 | 2017-12-08 | 南方医科大学 | The intensity modulated radiation therapy Forecasting Methodology of 3-dimensional dose distribution and its application in the works |
| EP3264298A1 (en) * | 2016-06-28 | 2018-01-03 | Ebit srl | Radiotherapy information system with treatment plan evaluation |
| CN107715314A (en) * | 2017-11-20 | 2018-02-23 | 徐榭 | Radiotherapy system and method based on deep learning |
| CN108367161A (en) * | 2017-06-05 | 2018-08-03 | 西安大医数码科技有限公司 | Radiotherapy system, data processing method and storage medium |
-
2018
- 2018-08-20 CN CN201810948783.4A patent/CN109166613A/en active Pending
Patent Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107072624A (en) * | 2014-06-18 | 2017-08-18 | Elekta股份有限公司 | system and method for automatic treatment plan |
| CN104225806A (en) * | 2014-08-28 | 2014-12-24 | 高献书 | Radiotherapy plan design method and system based on biologically effective dose |
| CN104815392A (en) * | 2015-04-10 | 2015-08-05 | 石峰 | Interactive radiotherapy plan system optimization method and system thereof |
| WO2017063691A1 (en) * | 2015-10-14 | 2017-04-20 | Brainlab Ag | Normal tissue sparing in radiation therapy treatment planning using prioritization for optimizing tissue indices |
| EP3264298A1 (en) * | 2016-06-28 | 2018-01-03 | Ebit srl | Radiotherapy information system with treatment plan evaluation |
| CN106902478A (en) * | 2017-02-22 | 2017-06-30 | 天津大学 | A kind of systematization radiotherapy biological effect appraisal procedure |
| CN108367161A (en) * | 2017-06-05 | 2018-08-03 | 西安大医数码科技有限公司 | Radiotherapy system, data processing method and storage medium |
| CN107441637A (en) * | 2017-08-30 | 2017-12-08 | 南方医科大学 | The intensity modulated radiation therapy Forecasting Methodology of 3-dimensional dose distribution and its application in the works |
| CN107715314A (en) * | 2017-11-20 | 2018-02-23 | 徐榭 | Radiotherapy system and method based on deep learning |
Cited By (22)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2020199836A1 (en) * | 2019-04-01 | 2020-10-08 | 安徽大学 | Method and system for dosimetry evaluation on organs at risk in esophageal radiotherapy plan |
| CN109859817A (en) * | 2019-04-01 | 2019-06-07 | 安徽大学 | Oesophagus radiotherapy treatment planning risk organ dose learns appraisal procedure, assessment system |
| CN109859817B (en) * | 2019-04-01 | 2022-12-09 | 安徽大学 | Method and system for evaluating risk organ dosimetry in esophageal radiotherapy plan |
| US11515025B2 (en) | 2019-04-01 | 2022-11-29 | The Affiliated Hospital Of Qingdao University | Dosimetry assessment method and system for organ at risk in esophageal radiotherapy plan |
| 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 |
| CN110197709A (en) * | 2019-05-29 | 2019-09-03 | 广州瑞多思医疗科技有限公司 | A kind of 3-dimensional dose prediction technique based on deep learning Yu priori plan |
| CN110232964B (en) * | 2019-06-04 | 2023-11-14 | 苏州雷泰智能科技有限公司 | Plan implementation method and device based on predicted dose guidance and Gaussian process optimization |
| CN110232964A (en) * | 2019-06-04 | 2019-09-13 | 苏州雷泰智能科技有限公司 | Plan implementation method and device based on predicted dose guidance and Gaussian process optimization |
| CN114007538B (en) * | 2019-06-27 | 2024-03-29 | 康坦手术股份有限公司 | Deep learning-based planning tissue ablation method |
| CN114007538A (en) * | 2019-06-27 | 2022-02-01 | 康坦手术股份有限公司 | Deep learning-based method for planning tissue ablation |
| JP2022530413A (en) * | 2019-09-26 | 2022-06-29 | ベイジン・センスタイム・テクノロジー・デベロップメント・カンパニー・リミテッド | Image processing methods and equipment, electronic devices, storage media and computer programs |
| CN110841205A (en) * | 2019-10-21 | 2020-02-28 | 温州医科大学附属第一医院 | Accurate dose verification method, device and equipment for tumor patient |
| CN110841205B (en) * | 2019-10-21 | 2021-06-22 | 温州医科大学附属第一医院 | Accurate dose verification device for tumor patients |
| WO2021090063A1 (en) * | 2019-11-08 | 2021-05-14 | Wang Qunxi | Cloud-based dose verification |
| CN111540437A (en) * | 2020-04-23 | 2020-08-14 | 北京大学第三医院（北京大学第三临床医学院） | Dose verification method and system based on artificial intelligence |
| CN116569190B (en) * | 2020-12-09 | 2024-04-30 | 光线搜索实验室公司 | Method and computer program product for dose map prediction for radiotherapy treatment planning |
| US11931599B2 (en) | 2020-12-09 | 2024-03-19 | Raysearch Laboratories Ab (Publ) | Method and computer program product for dose map prediction for radiotherapy treatment planning |
| CN116569190A (en) * | 2020-12-09 | 2023-08-08 | 光线搜索实验室公司 | Method and computer program product for dose map prediction for radiotherapy treatment planning |
| CN115223683B (en) * | 2022-08-19 | 2023-08-25 | 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） | Radiation treatment plan evaluation method, radiation treatment plan evaluation device, computer equipment and medium |
| CN115223683A (en) * | 2022-08-19 | 2022-10-21 | 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） | Radiotherapy plan evaluation method, radiotherapy plan evaluation device, computer equipment and medium |
| CN115862812B (en) * | 2023-02-28 | 2023-05-05 | 福建自贸试验区厦门片区Manteia数据科技有限公司 | Processing device for radiotherapy plan, storage medium and electronic equipment |
| CN115862812A (en) * | 2023-02-28 | 2023-03-28 | 福建自贸试验区厦门片区Manteia数据科技有限公司 | Radiotherapy plan processing device, storage medium and electronic device |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109166613A (en) | Radiotherapy treatment planning assessment system and method based on machine learning | |
| Hernandez et al. | What is plan quality in radiotherapy? The importance of evaluating dose metrics, complexity, and robustness of treatment plans | |
| US11386557B2 (en) | Systems and methods for segmentation of intra-patient medical images | |
| Fan et al. | Automatic treatment planning based on three‐dimensional dose distribution predicted from deep learning technique | |
| CN108717866A (en) | A kind of method, apparatus, equipment and the storage medium of the distribution of prediction radiotherapy planning dosage | |
| CN114423494B (en) | Radiotherapy treatment planning using a differentiable dose function | |
| CN107072624B (en) | System and method for automated treatment planning | |
| CN106651750A (en) | Method and system used for 2D/3D image registration based on convolutional neural network regression | |
| EP3300769B1 (en) | Radiation therapy planning optimization and visualization | |
| CN110841205B (en) | Accurate dose verification device for tumor patients | |
| CN107358607A (en) | Tumour radiotherapy visual monitoring and visual servo intelligent control method | |
| CN112581475B (en) | Method for predicting gamma passing rate of radiotherapy plan | |
| EP2994195B1 (en) | Interactive dose gradient based optimization technique to control imrt delivery complexity | |
| JP2017532092A (en) | System and method for segmenting medical images based on anatomical landmark-based features | |
| CN110085298A (en) | Intensity modulated radiation therapy plan 3-dimensional dose distribution forecasting method based on depth e-learning | |
| Wang et al. | Segmenting lung tumors on longitudinal imaging studies via a patient-specific adaptive convolutional neural network | |
| CN109069861A (en) | Dosage in proton therapy is distributed estimation | |
| CN104815392B (en) | A kind of interactive radiotherapy treatment planning system optimization system | |
| US10864382B2 (en) | System and method for radiation therapy treatment planning | |
| CN109562277A (en) | It is integrated in the automatic method and implementation system of the calibration of complicated radiotherapy dosage, reconstruction and the verifying in an environment | |
| CN112635024A (en) | Automatic planning and designing system for radiotherapy and construction method thereof | |
| CN105477789A (en) | Dynamic intensity-modulated radiotherapy method based on quadratic programming model suppressing total beam-out time | |
| Matsuura et al. | Predictive gamma passing rate of 3D detector array-based volumetric modulated arc therapy quality assurance for prostate cancer via deep learning | |
| Teng et al. | Beam-wise dose composition learning for head and neck cancer dose prediction in radiotherapy | |
| Pernes et al. | Fitting of superquadrics for breast modelling by geometric distance minimization |
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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190108 |
|
| RJ01 | Rejection of invention patent application after publication |