CN114298177A - Expansion enhancement method and system suitable for deep learning training data and readable storage medium - Google Patents
Expansion enhancement method and system suitable for deep learning training data and readable storage medium Download PDFInfo
- Publication number
- CN114298177A CN114298177A CN202111541880.XA CN202111541880A CN114298177A CN 114298177 A CN114298177 A CN 114298177A CN 202111541880 A CN202111541880 A CN 202111541880A CN 114298177 A CN114298177 A CN 114298177A
- Authority
- CN
- China
- Prior art keywords
- data
- training
- superposition
- augmentation
- original
- 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
- 238000012549 training Methods 0.000 title claims abstract description 107
- 238000000034 method Methods 0.000 title claims abstract description 23
- 238000013135 deep learning Methods 0.000 title claims abstract description 18
- 238000001228 spectrum Methods 0.000 claims abstract description 23
- 230000003416 augmentation Effects 0.000 claims description 10
- 238000004590 computer program Methods 0.000 claims 2
- 230000002708 enhancing effect Effects 0.000 abstract description 3
- 238000009826 distribution Methods 0.000 description 11
- 238000004364 calculation method Methods 0.000 description 9
- 238000005516 engineering process Methods 0.000 description 5
- 238000004422 calculation algorithm Methods 0.000 description 4
- 238000005520 cutting process Methods 0.000 description 4
- 238000013519 translation Methods 0.000 description 4
- 230000008569 process Effects 0.000 description 3
- 238000004980 dosimetry Methods 0.000 description 2
- 230000003993 interaction Effects 0.000 description 2
- 239000000463 material Substances 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 239000002245 particle Substances 0.000 description 2
- 238000012545 processing Methods 0.000 description 2
- 230000005855 radiation Effects 0.000 description 2
- 238000013528 artificial neural network Methods 0.000 description 1
- 230000003190 augmentative effect Effects 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 238000013527 convolutional neural network Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 230000004907 flux Effects 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 239000011159 matrix material Substances 0.000 description 1
- 230000005012 migration Effects 0.000 description 1
- 238000013508 migration Methods 0.000 description 1
- 230000002093 peripheral effect Effects 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 238000004088 simulation Methods 0.000 description 1
Images
Landscapes
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses an expansion enhancing method, a system and a readable storage medium suitable for deep learning training data, wherein the expansion enhancing method at least comprises the following steps: s1, acquiring original training data, classifying the original training data through a data classification module based on ray energy spectrum of the original training data, wherein the original training data are classified into the energy spectrum S with the same characteristicskThe original training data of (a) are classified into the same class; s2, training the module through the first data, based on the first random selection mode, from the energy spectrum S with the same characteristicskOriginal training ofSelecting N first data from the training data, wherein the N first data generate first new data in a first linear superposition mode; s3, selecting M second data from the first new data through a second data training module based on a second random selection mode, wherein the M second data generate second new data in a second linear superposition mode; s4, repeating steps S2 and S3 to generate training samples.
Description
Technical Field
The invention relates to the technical field of data enhancement, in particular to an expansion enhancement method and system suitable for deep learning training data and a readable storage medium.
Background
The greatest challenge in applying the neural network in the medical field is that a large amount of labeled data is lacked, and a large-scale training set (large-scale databases) covering all conditions as far as possible is adopted, so that the generalization capability (generralizability) of the model can be improved, and the model is prevented from being under-represented when new data is met. Similarly, it is difficult to traverse all the different intensity and shape dose distributions in the radiation treatment dose calculation. Therefore, the 'blank area' of the training set is made up by adopting an effective data enhancement technology, the trained model is ensured to have generalization capability, and the method is suitable for various conditions and is the key for influencing whether deep learning can fall on the ground in dose calculation.
Technology enhancement is used in deep learning training aiming at image processing in the prior art, and the technology comprises the following steps: 1) cutting (cropping). And randomly cutting the original data according to the size of the input data matrix to obtain data of different parts as new training data. 2) Scaling (scaling). The image is reduced or enlarged to generate new data, and each time the new data is generated, a random reduction and enlargement ratio is generated within a preset scaling range, and the original image is reduced or enlarged. 3) And (shift). And translating the image in the horizontal and vertical directions to generate new data, and randomly generating translation distances in the horizontal and vertical directions according to a preset translation distance range to generate new training data each time the new data is generated. 4) Rotation (rotation). The image is rotated to generate new data, and each time new data is generated, the image is randomly rotated by a certain angle within a preset rotation angle range by taking a certain set point as a rotation center to generate new data. 5) Flipping (flipping). The new data is generated by horizontally or vertically turning the image, and when the new data is generated, whether the turning is performed in the horizontal or vertical position is randomly determined, so that the new data is generated.
The prior art does not aim at the principle and characteristics of dosimetry, and effectively adds useful new types of data for dose calculation, and even new data which can generate errors, such as: 1) cutting, because the dose calculation at some point includes the contribution of the peripheral scattered rays, the cutting operation may cause the loss of part of the scattered information. 2) Scaling, particle interactions and migration are related to the actual physical distance, e.g. inverse square distance law, simple scaling without modification of the corresponding values may lead to errors in the partially reflected information. 3) Translation, rotation, and inversion, since the dose distribution generated is fixed when the relative position of the particles and the medium distribution is fixed, simple translation, rotation, and inversion operations do not generate new data, e.g., do not cover generating new different energy spectra, doses at different beam intensities and shapes, and do not generate substantially new data. The present invention therefore aims to provide a data enhancement method which overcomes the above-mentioned drawbacks.
Disclosure of Invention
The invention aims to overcome the defects of the prior art and provides an expansion enhancing method, system and readable storage medium suitable for deep learning training data.
The purpose of the invention is realized by the following technical scheme: an augmentation and enhancement method suitable for deep learning training data, which is applied to a data training server in communication connection with a plurality of user equipment, and at least comprises the following steps: s1, acquiring original training data, classifying the original training data through a data classification module based on ray energy spectrum of the original training data, wherein the original training data are classified into the energy spectrum S with the same characteristicskThe original training data of (a) are classified into the same class; s2, training the module through the first data, based on the first random selection mode, from the energy spectrum S with the same characteristicskSelecting N first data from the original training data, wherein the N first data generate first new data in a first linear superposition mode; s3, selecting M second data from the first new data through a second data training module based on a second random selection mode, wherein the M second data are stacked according to a second linear modeGenerating second new data in an adding mode; s4, repeating steps S2 and S3 to generate training samples.
Preferably, the first linear superposition mode at least comprises the following steps: s201, respectively naming the N randomly extracted data as (T)1,k,D1,k),(T2,k,D2,k)……(TN,k,DN,k) (ii) a S202, the N data are linearly overlapped based on a first overlapping formula to obtain the first new data (T'k,D’k) The first superposition formula isWherein T is input data of the training set, D is target data of the training set, ciIs a first superposition proportion ciIs a random number between 0 and 1, i is a training step number, and the first superposition proportion ciSatisfy the requirement of
Preferably, the second linear superposition mode at least comprises the following steps: s301, the randomly extracted M data are named as (T'1,D’1),(T’2,D’2)……(T’M,D’M) (ii) a S302, the M data are linearly overlapped based on a second overlapping formula to obtain the second new data (T ', D'), and the second overlapping formula isWherein d iskIs a second superposition proportion dkIs a random number between 0 and 1, k is a training step number, and the second superposition proportion dkSatisfy the requirement of
Preferably, the number N of the first data is smaller than SkClassifying the total amount of original data in the data, the amount M of the second data being less than SkAnd (5) classifying the categories.
Preferably, in the case of repeating steps S2 and S3 to generate training samples, the number N of the first data and the number M of the second data are configured to be randomly set, and the first superposition ratio c isiAnd a second superposition ratio dkAre configured to be randomly arranged.
The invention has the following advantages:
the invention relates to an enhancement technology applied to dose calculation aiming at deep learning, which carries out classification and superposition on training data in a training process according to the principle of dosimetry and the characteristic of linear superposition and according to ray energy, ray intensity and shape to obtain the effect of effectively and randomly generating the training data with new characteristics, further effectively making up the blank area of an original training set, realizing the purposes of improving the generalization capability of a training model, processing the new data and avoiding overfitting.
Drawings
FIG. 1 is a flow chart illustrating an augmentation method for deep learning training data according to the present invention;
fig. 2 is a schematic view of a modular structure of the augmentation system of the present invention.
In the figure, 1-data training server, 2-user equipment, 3-data classification module, 4-first data training module, and 5-second data training module.
Detailed Description
The invention will be further described with reference to the accompanying drawings, without limiting the scope of the invention to the following:
as shown in fig. 1, the present application provides an augmentation method suitable for deep learning training data, which at least includes the following steps:
s1, acquiring original training data, and classifying the original training data based on ray energy spectrum of the original training data, wherein the original training data are classified to have the same characteristic energy spectrum SkThe original training data of (a) are classified into the same class;
specifically, the original training data is obtained according to the following steps:
s101, converting the image information into a medium material and an electron density distribution diagram. The image is a CT image, and HU values of the CT image are converted into an electron density distribution image according to an HU-electron density conversion curve of a machine for acquiring the CT image.
S102, calculating the amount TERMA of the interaction between the initial incident photons and the medium according to the beam condition, the material of the medium provided by the image and the electron density distribution. This implementation calculates for the monoenergetic, the slave source for the photon ray with monoenergetic energy ETo the point of computationThe formula of TERMA isIn the formulaAndare respectively in the gridAndthe attenuation coefficient of photons (attenuation coefficient) of (a), which is related to photon energy and medium, mass attenuation coefficient obtained from National Institute of Standards and Technology (NIST) queries,is composed ofThe density of the medium at the point (b),is the beam flux distribution.
And S103, training target data. The training target data is obtained by adopting Monte Carlo dose algorithm calculation, and is paired with the input data during calculation and generated under the same CT and the same ray energy and distribution.
And S104, training data generation conditions. The training data was generated under the following conditions: (1) the ray energy is generated by randomly drawing one energy spectrum from the beam energy spectrums of three medical linear accelerators of 4MV, 10MV and 24 MV. (2) By randomly setting the field conditions, the field range is in the crossline direction, and X1: -20cm to 15cm, X2: -15cm to 20cm, inline orientation, Y1: -20cm to 10cm, Y2: -10cm to 20 cm. The isocenter is randomly located within the phantom. And calculating the corresponding TERMA distribution and the dose distribution calculated by the target dose algorithm according to the radiation field condition and the electron density distribution.
Preferably, having the same characteristic spectrum SkThe classification of the raw training data into the same class can be generated as follows: according to the ray energy spectrum 4MV, 10MV and 24MV of the generated original training data, the original training data is divided into three classes, and S is used respectively1,S2,S3And (4) performing representation.
S2, based on the first random selection mode, selecting the energy spectrum S with the same characteristicskN first data are selected from the original training data, and the N first data generate first new data according to a first linear superposition mode.
Specifically, for original training data with the same ray energy spectrum, 5 first data (T) are randomly extracted1,k,D1,k),(T2,k,D2,k)……(T5,k,D5,k) Generating a first new data (T ') in a first linear superposition'k,D’k). Generating 1 first new data per energy spectrum, 3 first new data in total, and respectively using (T'1,D’1),(T’2,D’2),(T’3,D’3) And (4) showing.
Preferably, the first linear superposition mode at least comprises the following steps:
s201, respectively naming the N randomly extracted data as (T)1,k,D1,k),(T2,k,D2,k)……(TN,k,DN,k);
S202, the N data are linearly overlapped based on a first overlapping formula to obtain the first new data (T'k,D’k) The first superposition formula isWherein T is input data of the training set, D is target data of the training set, ciIs a first superposition proportion ciIs a random number between 0 and 1, i is a training step number, and the first superposition proportion ciSatisfy the requirement of
S3, M second data are selected from the first new data based on a second random selection mode, and the M second data generate second new data according to a second linear superposition mode.
Specifically, the first new data (T ') is generated from the newly generated first new data'k,D’k) In (1), 3 pieces of second data (T'1,D’1),(T’2,D’2),(T’3,D’3) The three second data generate second new data (T ", D") in a second linear superposition.
Preferably, the second linear superposition mode at least comprises the following steps:
s301, the randomly extracted M data are named as (T'1,D’1),(T’2,D’2)……(T’M,D’M);
S302, the M data are linearly overlapped based on a second overlapping formula to obtain the second new data (T ', D'), and the second overlapping formula isWherein d iskIs a second superposition proportion dk0 to 1Random number of cells, k being the number of training steps, the second superposition ratio dkSatisfy the requirement of
S4, repeating steps S2 and S3 to generate training samples.
Specifically, the generated second new data (T ", D") is sent to the network training process, and the steps S2 and S3 are repeated to continue generating new data until the training is finished.
Preferably, the number N of the first data is smaller than SkClassifying the total amount of original data in the data, the amount M of the second data being less than SkAnd (5) classifying the categories. In the case of repeating steps S2 and S3 to generate training samples, the number N of the first data and the number M of the second data are configured to be randomly set, the first superposition ratio ciAnd a second superposition ratio dkAre configured to be randomly arranged.
Preferably, the enhancement technique for the training data in the training process can be applied to model training including, but not limited to, convolutional neural network framework, etc. The different incident ray energy spectrums include but are not limited to beam-out energy spectrums of different accelerating voltages of the existing medical linear accelerator, manually fitted energy spectrums with different characteristics, single energy spectrums and the like. The dose calculation results as training targets can be obtained by any treatment planning system or software with dose calculation functions, for example, by pencil beam algorithm, cartridge string convolution algorithm, Monte Care simulation, etc.
Preferably, as shown in fig. 2, the present application further provides an augmented enhancement system suitable for deep learning training data, which includes at least a data training server 1 and a plurality of user devices 2. The data training server 1 is used to provide raw training data. The data training server is able to transmit training data that has been trained to the user device 2 for application by the user device 2. The augmentation enhancement system further comprises a data classification module 3, a first data training module 4 and a second data training module 5. The data classification module 3 is configured to execute step S1. The first data training module 4 is configured to execute step S2. The third data training module 5 is configured to execute step S3.
Finally, it should be noted that: although the present invention has been described in detail with reference to the foregoing embodiments, it will be apparent to those skilled in the art that changes may be made in the embodiments and/or equivalents thereof without departing from the spirit and scope of the invention. Any modification, equivalent replacement, or improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.
Claims (7)
1. An augmentation method suitable for deep learning training data, which is applied to a data training server communicatively connected to a plurality of user devices, the augmentation method at least includes the following steps:
s1, acquiring original training data, classifying the original training data through a data classification module (3) based on ray energy spectrum of the original training data, wherein the original training data are classified into energy spectrums S with the same characteristicskThe original training data of (a) are classified into the same class;
s2, training the module (4) through the first data, and based on the first random selection mode, selecting the energy spectrum S with the same characteristicskSelecting N first data from the original training data, wherein the N first data generate first new data in a first linear superposition mode;
s3, selecting M second data from the first new data through a second data training module (5) based on a second random selection mode, wherein the M second data generate second new data in a second linear superposition mode;
s4, repeating steps S2 and S3 to generate training samples.
2. The augmentation-enhancing method for deep-learning training data according to claim 1, wherein the first linear superposition mode at least comprises the following steps:
s201, respectively naming the N randomly extracted data as (T)1,k,D1,k),(T2,k,D2,k)……(TN,k,DN,k);
S202, the N data are linearly overlapped based on a first overlapping formula to obtain the first new data (T'k,D’k) The first superposition formula isWherein T is input data of the training set, D is target data of the training set, ciIs a first superposition proportion ciIs a random number between 0 and 1, i is a training step number, and the first superposition proportion ciSatisfy the requirement of
3. The augmentation-enhancing method for deep-learning training data according to claim 2, wherein the second linear superposition mode at least comprises the following steps:
s301, the randomly extracted M data are named as (T'1,D’1),(T’2,D’2)……(T’M,D’M);
S302, the M data are linearly overlapped based on a second overlapping formula to obtain the second new data (T ', D'), and the second overlapping formula isWherein d iskIs a second superposition proportion dkIs a random number between 0 and 1, k is a training step number, and the second superposition proportion dkSatisfy the requirement of
4. The augmentation-enhancing method for deep-learning training data as claimed in claim 1, wherein the first stepThe number N of data is less than SkClassifying the total amount of original data in the data, the amount M of the second data being less than SkAnd (5) classifying the categories.
5. The augmentation-enhancing method for deep-learning training data according to claim 3, wherein in the case of repeating the steps S2 and S3 to generate training samples, the number N of the first data and the number M of the second data are configured to be randomly set, and the first superposition ratio c isiAnd a second superposition ratio dkAre configured to be randomly arranged.
6. An augmentation system suitable for deep learning training data, comprising a data training server and a plurality of user devices communicatively connected to the data training server, wherein the data training server is configured to perform the augmentation method of any one of claims 1 to 5.
7. A computer-readable storage medium, in which a computer program is stored, which computer program is run to perform the augmentation enhancement method of any one of claims 1 to 5.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202111541880.XA CN114298177A (en) | 2021-12-16 | 2021-12-16 | Expansion enhancement method and system suitable for deep learning training data and readable storage medium |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202111541880.XA CN114298177A (en) | 2021-12-16 | 2021-12-16 | Expansion enhancement method and system suitable for deep learning training data and readable storage medium |
Publications (1)
Publication Number | Publication Date |
---|---|
CN114298177A true CN114298177A (en) | 2022-04-08 |
Family
ID=80967864
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202111541880.XA Pending CN114298177A (en) | 2021-12-16 | 2021-12-16 | Expansion enhancement method and system suitable for deep learning training data and readable storage medium |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN114298177A (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN115270588A (en) * | 2022-08-03 | 2022-11-01 | 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) | Fluence spectral distribution calculation modeling and fluence spectral distribution calculation method and device |
Citations (13)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107392130A (en) * | 2017-07-13 | 2017-11-24 | 西安电子科技大学 | Classification of Multispectral Images method based on threshold adaptive and convolutional neural networks |
CN109872332A (en) * | 2019-01-31 | 2019-06-11 | 广州瑞多思医疗科技有限公司 | A kind of 3 d medical images method for registering based on U-NET neural network |
CN110232396A (en) * | 2019-04-09 | 2019-09-13 | 贵州大学 | X-ray breast image deep learning classification method |
US20190370969A1 (en) * | 2018-05-30 | 2019-12-05 | Siemens Healthcare Gmbh | Methods for generating synthetic training data and for training deep learning algorithms for tumor lesion characterization, method and system for tumor lesion characterization, computer program and electronically readable storage medium |
CN110956634A (en) * | 2020-02-26 | 2020-04-03 | 南京慧脑云计算有限公司 | Deep learning-based automatic detection method and system for cerebral microhemorrhage |
CN111239791A (en) * | 2020-01-17 | 2020-06-05 | 卡迪诺科技(北京)有限公司 | Radiation dose field simulation method and device |
CN111275691A (en) * | 2020-01-22 | 2020-06-12 | 北京邮电大学 | Small sample tumor necrosis rate classification prediction device based on deep learning |
CN111291833A (en) * | 2020-03-20 | 2020-06-16 | 京东方科技集团股份有限公司 | Data enhancement method and data enhancement device applied to supervised learning system training |
CN111814607A (en) * | 2020-06-24 | 2020-10-23 | 中国地质大学(武汉) | Deep learning model suitable for small sample hyperspectral image classification |
WO2020236206A1 (en) * | 2019-05-23 | 2020-11-26 | Google Llc | Methods, systems, and media for relighting images using predicted deep reflectance fields |
WO2021057174A1 (en) * | 2019-09-26 | 2021-04-01 | 北京市商汤科技开发有限公司 | Image processing method and apparatus, electronic device, storage medium, and computer program |
CN112949738A (en) * | 2021-03-17 | 2021-06-11 | 西安电子科技大学 | Multi-class unbalanced hyperspectral image classification method based on EECNN algorithm |
US20210383148A1 (en) * | 2020-06-09 | 2021-12-09 | Shandong University | Automatic data enhancement expansion method, recognition method and system for deep learning |
-
2021
- 2021-12-16 CN CN202111541880.XA patent/CN114298177A/en active Pending
Patent Citations (13)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107392130A (en) * | 2017-07-13 | 2017-11-24 | 西安电子科技大学 | Classification of Multispectral Images method based on threshold adaptive and convolutional neural networks |
US20190370969A1 (en) * | 2018-05-30 | 2019-12-05 | Siemens Healthcare Gmbh | Methods for generating synthetic training data and for training deep learning algorithms for tumor lesion characterization, method and system for tumor lesion characterization, computer program and electronically readable storage medium |
CN109872332A (en) * | 2019-01-31 | 2019-06-11 | 广州瑞多思医疗科技有限公司 | A kind of 3 d medical images method for registering based on U-NET neural network |
CN110232396A (en) * | 2019-04-09 | 2019-09-13 | 贵州大学 | X-ray breast image deep learning classification method |
WO2020236206A1 (en) * | 2019-05-23 | 2020-11-26 | Google Llc | Methods, systems, and media for relighting images using predicted deep reflectance fields |
WO2021057174A1 (en) * | 2019-09-26 | 2021-04-01 | 北京市商汤科技开发有限公司 | Image processing method and apparatus, electronic device, storage medium, and computer program |
CN111239791A (en) * | 2020-01-17 | 2020-06-05 | 卡迪诺科技(北京)有限公司 | Radiation dose field simulation method and device |
CN111275691A (en) * | 2020-01-22 | 2020-06-12 | 北京邮电大学 | Small sample tumor necrosis rate classification prediction device based on deep learning |
CN110956634A (en) * | 2020-02-26 | 2020-04-03 | 南京慧脑云计算有限公司 | Deep learning-based automatic detection method and system for cerebral microhemorrhage |
CN111291833A (en) * | 2020-03-20 | 2020-06-16 | 京东方科技集团股份有限公司 | Data enhancement method and data enhancement device applied to supervised learning system training |
US20210383148A1 (en) * | 2020-06-09 | 2021-12-09 | Shandong University | Automatic data enhancement expansion method, recognition method and system for deep learning |
CN111814607A (en) * | 2020-06-24 | 2020-10-23 | 中国地质大学(武汉) | Deep learning model suitable for small sample hyperspectral image classification |
CN112949738A (en) * | 2021-03-17 | 2021-06-11 | 西安电子科技大学 | Multi-class unbalanced hyperspectral image classification method based on EECNN algorithm |
Non-Patent Citations (1)
Title |
---|
唐校辉;廖欣;陈雷霆;陈文昭;: "基于改进Tri-Training算法的健康大数据分类模型研究", 现代计算机(专业版), no. 20, 15 July 2017 (2017-07-15) * |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN115270588A (en) * | 2022-08-03 | 2022-11-01 | 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) | Fluence spectral distribution calculation modeling and fluence spectral distribution calculation method and device |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CA2557888C (en) | Computation of radiating particle and wave distributions using a generalized discrete field constructed from representative ray sets | |
Gifford et al. | Comparison of a finite-element multigroup discrete-ordinates code with Monte Carlo for radiotherapy calculations | |
US20230302296A1 (en) | Method, apparatus, and system for simulating a particle transport and determining human dose in a radiotherapy | |
Famulari et al. | RapidBrachyMCTPS: a Monte Carlo-based treatment planning system for brachytherapy applications | |
EP3566747A1 (en) | Medical image data-based method for establishing smooth geometric model | |
CN114298177A (en) | Expansion enhancement method and system suitable for deep learning training data and readable storage medium | |
Mackeprang et al. | Independent Monte-Carlo dose calculation for MLC based CyberKnife radiotherapy | |
Zhu et al. | A preliminary study of a photon dose calculation algorithm using a convolutional neural network | |
Malusek et al. | A model‐based iterative reconstruction algorithm DIRA using patient‐specific tissue classification via DECT for improved quantitative CT in dose planning | |
Bahreyni-Toosi et al. | Monte Carlo simulation of a 6 MV X-ray beam for open and wedge radiation fields, using GATE code | |
Babcock et al. | Increasing the speed of DOSXYZnrc Monte Carlo simulations through the introduction of nonvoxelated geometries | |
Grzanka | Modelling beam transport and biological effectiveness to develop treatment planning for ion beam radiotherapy | |
CN113426030B (en) | Proton dosage calculation method and device | |
Vandervoort et al. | Evaluation of a new commercial Monte Carlo dose calculation algorithm for electron beams | |
Shaw | Evaluation of the effects of bolus air gaps on surface dose in radiation therapy and possible clinical implications | |
Liu et al. | Fast Monte Carlo simulation for total body irradiation using a teletherapy unit | |
Solovyev et al. | Comparative analysis of MCNPX and GEANT4 codes for fast-neutron radiation treatment planning | |
Yuan et al. | Deterministic photon kerma distribution based on the Boltzmann equation for external beam radiation therapy | |
CN107441635B (en) | It is a kind of for reducing the multiple tumor radioactive ray method for optimizing route for the treatment of time | |
CN116561834A (en) | Method and system for optimally designing neutron source target structural parameters of electron accelerator | |
Fix et al. | Auto-commissioning of a Monte Carlo electron beam model with application to photon MLC shaped electron fields | |
CN107185117B (en) | Sigmatron zone routing optimization method towards multiple tumor radiotherapy | |
Battista | Conceptual Overview of Algorithms | |
Zhang et al. | Development of multilevel mesh adaptivity for global variance reduction in Monte Carlo code cosRMC | |
Quarz et al. | Deep learning-based voxel sampling for particle therapy treatment planning |
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 |