CN109966662B

CN109966662B - System for verifying radiation therapy dosage

Info

Publication number: CN109966662B
Application number: CN201910360026.XA
Authority: CN
Inventors: 王卫东; 闫梦梦
Original assignee: Sichuan Cancer Hospital
Current assignee: Sichuan Cancer Hospital
Priority date: 2019-04-30
Filing date: 2019-04-30
Publication date: 2020-11-24
Anticipated expiration: 2039-04-30
Also published as: CN109966662A

Abstract

The invention relates to a system for verifying radiation therapy dosage, comprising the following steps: s1, receiving a CT image before radiotherapy, and delineating a target area in the CT image; s2, extracting a plurality of image omics characteristics of the target area; s3, screening the imaging omics characteristics which have obvious correlation with the tumor regression ratio TRR from the extracted imaging omics characteristics; and S4, setting a dose value, identifying the CT image processed in the step S3 by using a pre-trained radiologic model, and outputting to obtain the tumor regression proportion TRR under the given dose value. The invention provides a new idea, a radiology model is obtained based on machine learning method training, a given metering value is verified by using TRR output by the radiology model, a prediction result is not influenced by single dose, and a verification result is more reliable.

Description

System for verifying radiation therapy dosage

Technical Field

The invention relates to the technical field of medical treatment, in particular to a system for verifying radiation treatment dosage.

Background

The radiotherapy dose is adjusted according to the individual biological effect dose (Bd), and the formula is Bd-E/alpha-nd (1+ d/[ alpha/beta ])])-log_e2(T-TK)/α Tp, wherein Bd is BED, is the biologically effective dose, E is the total exponential killing of the cells, α is the irreparable killing coefficient, i.e., the exponential killing per Gy of irradiation, β is the repairable killing coefficient, i.e., the square of the exponential killing per Gy of irradiation, n is the number of fractionated irradiation, d is the single fraction dose, T is the total irradiation time, TK is the cell re-proliferation start time, and Tp is the average time for doubling the number of cells in successive irradiations. Reference may be made to "The linear quadratic form and progression in fractionated radiotherapy", authors: fowler, j.f., BrJ Radiology 1989; 62:1261-1269.

In radiation therapy, to ensure that a radiation Plan is accurately delivered to a patient, the dose distribution calculated using the Treatment Planning System (TPS) must be confirmed by experimental data. The reduced fit of the traditional "L-Q model" and its derived Biological Effect Dose (BED) model in the high dose region may overestimate tumor control and underestimate the response to normal tissue. The L-Q model only fits data through in vitro experiments, departs from clinical practice, does not reflect vascular and interstitial injuries, ignores the influence of radiation-resistant cell subsets on overall injuries, and is not suitable for equivalent biological dose evaluation of single large-dose radiotherapy. The cell damage mechanism shows a linear-quadratic relation, and the L-Q model is suitable for equivalent biological dose conversion under the condition that the single dose does not exceed 15-20 Gy.

Disclosure of Invention

It is an object of the present invention to provide a method and system for verifying radiation treatment dose, not only as a new verification method, but also to adjust the dose according to the outputted TRR.

To this end, embodiments of the present invention provide a method of verifying a radiation therapy dose, comprising the steps of:

s1, receiving a CT image before radiotherapy, and delineating a target area in the CT image;

s2, extracting a plurality of image omics characteristics of the target area;

s3, screening the imaging omics characteristics which have obvious correlation with the tumor regression ratio TRR from the extracted imaging omics characteristics;

and S4, setting a dose value, identifying the CT image processed in the step S3 by using a pre-trained radiologic model, and outputting to obtain the tumor regression proportion TRR under the given dose value.

In one embodiment, the step of screening for a proteomic signature having a significant correlation with the tumor regression rate TRR comprises:

s31, rewriting the outliers in the extracted plurality of proteomic features into the maximum or minimum values of the plurality of proteomic features excluding the outliers;

s32, carrying out Z-score normalization processing on all the image omics characteristic values processed by the S31;

and S33, selecting the data value after the Z-score normalization treatment according to the spearman correlation coefficient, and selecting the characteristic with obvious correlation with the tumor regression rate.

As an embodiment, the radiology model is trained by the following steps:

calculating a tumor regression proportion TRR according to the CT images after radiation treatment before radiation treatment and under different doses, and marking the CT images before radiation treatment according to the TRR; v ═ Σ_iV_i，

Wherein, V_iIs the volume of voxel i in the target regionV1 is the pre-radiotherapy target volume and V2 is the post-radiotherapy target volume;

for CT images before radiotherapy, extracting the imaging group characteristics with significant correlation with TRR according to the steps S2-S3;

and (3) learning and predicting the marked CT image extracted with the characteristics which have obvious correlation with the TRR by using a bag algorithm, evaluating the prediction result, stopping training if the accuracy of the prediction result is greater than a set threshold value and/or the area AUC under the curve is greater than a set value, otherwise, updating the parameters, and circularly executing the step.

In this embodiment, a system for verifying radiation therapy dosage is also provided, including:

the target area delineation module is used for receiving the CT image before radiotherapy and delineating a target area in the CT image;

the characteristic extraction module is used for extracting a plurality of image omics characteristics of the target area;

the characteristic screening module is used for screening the image omics characteristics which have obvious correlation with the tumor regression ratio TRR from the extracted image omics characteristics;

and the TRR prediction module is used for giving a dose value, identifying the screened CT images with the imaging omics characteristics which have obvious correlation with the TRR by utilizing a pre-trained radiolomics model, and outputting to obtain the tumor regression proportion TRR under the given dose value.

In another aspect, the present invention also provides a computer-readable storage medium comprising computer-readable instructions, which, when executed, cause a processor to perform the operations of the method according to any one of the embodiments of the present invention.

In another aspect, the present invention also provides an electronic device, including: a memory storing program instructions; and the processor is connected with the memory and executes the program instructions in the memory to realize the steps of the method in any embodiment of the invention.

Compared with the prior art, the method or the system provides a new thought, a radiologic model is obtained based on machine learning method training, the given metering value is verified by using TRR output by the radiologic model, the prediction result is not influenced by single dose, the model is obtained based on actual CT image training, vascular and interstitial injuries and the influence of radiation-resistant cell subgroups on the overall injury are reflected, and the verification result is more reliable.

Drawings

In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings needed to be used in the embodiments will be briefly described below, it should be understood that the following drawings only illustrate some embodiments of the present invention and therefore should not be considered as limiting the scope, and for those skilled in the art, other related drawings can be obtained according to the drawings without inventive efforts.

FIG. 1 is a flow chart of a method of verifying a radiation therapy dose as described in the present embodiment.

Figure 2 is a flow chart of the selection of the iconomics features with significant correlation to Tumor Regression Ratio (TRR) as described in this example.

Fig. 3 is a flowchart of training a radiology model according to the present embodiment.

Fig. 4 is a functional block diagram of a system for verifying radiation treatment dosage as described in the present embodiment.

FIG. 5 is a functional block diagram of the electronic device described in the embodiments.

Detailed Description

The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. The components of embodiments of the present invention generally described and illustrated in the figures herein may be arranged and designed in a wide variety of different configurations. Thus, the following detailed description of the embodiments of the present invention, presented in the figures, is not intended to limit the scope of the invention, as claimed, but is merely representative of selected embodiments of the invention. All other embodiments, which can be derived by a person skilled in the art from the embodiments of the present invention without making any creative effort, shall fall within the protection scope of the present invention.

Referring to fig. 1, the present embodiment provides a method for verifying a radiation therapy dose, including the following steps:

s1, receiving the input pre-radiotherapy CT (Computed Tomography) image, and delineating a target Volume (Gross Tumor Volume) in the CT image. The target area can be drawn manually.

And S2, extracting the imaging omics characteristics of the target area. In this embodiment, 57 image omics features are automatically extracted by LIFEx software, which includes: first order features such as gray scale and shape; texture features such as Grey-Level Zone Length Matrix, Grey-Level Run Length Matrix, Neighborhod Grey-Level differential Matrix, and Grey Level Co-occurence Matrix. The 57 image omics characteristics are not described in detail here, and can be automatically extracted by LIFEx software in actual operation, or can be referred to related technical documents, for example, refer to "LIFEx, a free for radio presentation in multi-modal imaging to access updates in the characterization of the Cancer Research, 2018; 4786-. Also for example, LIFEx-Texture, MSoussan, F Orlhac, M Boubaya, L Zelek, M Ziol, V Eder, I Buvat; relationship between molecular biology measured on FDG-PET/CT and clinical diagnostic factors in innovative scientific cancer. plos One 9: e94017,2014.

S3, selecting a proteomic feature having a significant correlation with the Tumor Regression Ratio (TRR) from the 57 extracted proteomic features.

Specifically, referring to fig. 2, the process of selecting a proteomic signature with a significant correlation to Tumor Regression Ratio (TRR) comprises the following steps:

s31, scaling the outliers, that is, changing the outliers of the 57 feature values to the maximum or minimum values of the 57 feature values except the outliers, so as to eliminate the influence of the extreme values on the features. Outliers (outliers), also known as outliers, mean that one or several values in the data differ significantly from other values. As an example, a value having a probability of deviating from the observation mean value of 1/(2n) or less, where n is the number of cases, may be generally defined as an outlier. Of course, the user can also customize the device according to the needs. For example, if there is 1,2, or 10 in a group of data, 10 is considered as an outlier, and the value is modified to 1 or 2.

S32, Z-score normalization processing. That is, the actual data value is converted to a new data representation by the following equation: z ═ x- μ)/σ, where x is a specific eigenvalue, μ is the average of the eigenvalues, and σ is the standard deviation.

For example, for a sequence: 2, 5, 8, 13, 18, normalized by Z-score to: -1.12859, -.65834, -.18810,0.59564,1.37938.

And S33, selecting the characteristics with significant correlation with the Tumor Regression Rate (TRR) according to the spearman correlation coefficient, carrying out a two-tail test, namely carrying out correlation detection on each characteristic value and the Tumor Regression Rate (TRR), and judging the characteristics with significant correlation with the TRR by using the image group characteristics with the significance level smaller than a set value (for example, 0.05). The spearman correlation assay is known to be a mature technique, so the specific assay procedure is not described here in detail.

The last 5 selected characteristics are respectively as follows: HISTO _ Skewness, skewed histogram; HISTO _ Energy, histogram of Energy; SHAPE _ Volume, (Volume); GLRLM _ LRE (The gray-level Run length matrix-Long Run Emphasis) gray level Run matrix-length running point; GLZLM _ SZE, gray zone length matrix _ short zone emphasis.

And S4, identifying the CT image processed in the step S3 by utilizing a pre-trained radiologic model, giving a dosage value, outputting the tumor regression proportion (TRR) under the dosage value, and verifying whether the given dosage value is reasonable or not according to the output TRR.

The radiologic model has the function of outputting the corresponding TRR when the set dose is input by learning the actual CT images under different doses in advance, so that the dose can be adjusted according to the output TRR, the BED can be adjusted by adjusting the dose, the corresponding TRR is output, and the dose required by complete regression of the tumor volume can be determined according to the output TRR.

In this embodiment, the radiology model is obtained by training through the following method:

step 1, calculating a tumor regression proportion TRR according to CT images after radiation treatment before and after radiation treatment and under different doses (the CT images of the same patient are obtained before and after the radiation treatment), and marking the CT images before the radiation treatment according to the following Table2, namely marking the CT images with TRR of 0-0.6 as 1 and marking the CT images with TRR of 0.6-1 as 2. The tumor regression ratio TRR was calculated in the following manner: v ═ Σ_iV_i，

Wherein, V_iIs the volume of voxel i within the target volume, V1 is the pre-radiotherapy target volume, and V2 is the post-radiotherapy target volume.

And 2, extracting 5 image omics characteristics which are obviously related to TRR from the CT image before radiotherapy according to the methods of the steps S2-S3. In training, in order to make the radiation dose comparable between cases, the biological effect dose (Bd) is introduced as a non-imaging omics feature into the feature set. The feature set description is shown as Table 3.

There is no sequential division between step 1 and step 2 above.

And 3, learning and predicting the marked CT image with the extracted characteristic which is obviously related to the TRR by using a bag algorithm, so that the corresponding TRR under the given dose can be output by the radiology group model. The bagging algorithm has the advantages that the accuracy is obviously higher than that of any single classifier in the combination, the performance of the classifier on larger noise is not poor, the robustness is realized, and overfitting is not easy to occur, so the bagging algorithm is adopted in the embodiment.

Specifically, referring to fig. 3, a plurality of training sets D1-Dk are generated based on a learning data set D, then each training set is trained and predicted by using a corresponding number of classifiers M1-Mk, then a new testing data set is input into different classifiers to obtain K different prediction results, and finally the prediction result with the largest number (i.e., a small number subject to majority) is used as the final output result. For example, if 55 results obtained by the 99-classification method are 1 and 44 results are 0, the final result is determined to be 1. Therefore, K is preferably an odd number.

In one training example, the study dataset is CT images of 34 lung cancer patients (both processed through step 1 and step 2), and the clinical features are shown in Table 1. Let TRRv be a prediction label, and be classified into two levels, as shown in Table 2.

Table 1.

Table 2.

TRR	(-∞,0.6)	[0.6,1)
			TRR_v	1	2

Table 3.

And 4, evaluating the output result of the step 3, stopping training if the accuracy is greater than a set threshold value and/or the area under the curve AUC is greater than the set threshold value, otherwise, returning to the step 3, updating the model parameters, and circularly executing the step 3-4.

The prediction model uses a bag-packed algorithm in ensemble learning, and is cross-validated by 10 times. The algorithm showed excellent results with an accuracy of 97.1% and an area under the curve (AUC) of 0.99.

Based on the same inventive concept as the above method, the present embodiment also provides a system for verifying radiation therapy dosage. As shown in fig. 4, the system for verifying radiation therapy dosage comprises:

Wherein, the characteristic screening module includes:

the data scaling submodule is used for rewriting the outliers in the extracted multiple image omics characteristics into the maximum value or the minimum value except for the outliers in the multiple image omics characteristics;

the normalization submodule is used for carrying out Z-score normalization processing on all the image omics characteristic values output by the data scaling submodule;

and the correlation detection submodule is used for selecting the characteristic with obvious correlation with the tumor regression proportion according to the spearman correlation coefficient for the data value output by the normalization submodule.

Wherein, the radiology model is obtained by training the following steps:

Wherein, V_iIs the volume of voxel i within the target volume, V1 is the pre-radiotherapy target volume, V2 is the post-radiotherapy target volume;

extracting an image omics characteristic which is obviously related to TRR from the CT image before radiotherapy;

As shown in fig. 5, the present embodiment also provides an electronic device, which may include a processor 51 and a memory 52, wherein the memory 52 is coupled to the processor 51. It is noted that the figure is exemplary and that other types of structures may be used in addition to or in place of the structure to implement data extraction, chart redrawing, communication, or other functionality.

As shown in fig. 5, the electronic device may further include: an input unit 53, a display unit 54, and a power supply 55. It is to be noted that the electronic device does not necessarily have to comprise all the components shown in fig. 5. Furthermore, the electronic device may also comprise components not shown in fig. 5, reference being made to the prior art.

The processor 51, also sometimes referred to as a controller or operational control, may comprise a microprocessor or other processor device and/or logic device, the processor 51 receiving input and controlling operation of the various components of the electronic device.

The memory 52 may be one or more of a buffer, a flash memory, a hard drive, a removable medium, a volatile memory, a non-volatile memory, or other suitable devices, and may store the configuration information of the processor 51, the instructions executed by the processor 51, the recorded table data, and other information. The processor 51 may execute a program stored in the memory 52 to realize information storage or processing, or the like. In one embodiment, a buffer memory, i.e., a buffer, is also included in the memory 52 to store the intermediate information.

The input unit 53 is used, for example, to provide the processor 51 with pre-or post-radiation CT images, or with given dose values. The display unit 54 is used for displaying the output TRR, or various intermediate result maps in the process, and may be, for example, an LCD display, but the present invention is not limited thereto. The power supply 55 is used to provide power to the electronic device.

Embodiments of the present invention further provide a computer readable instruction, where when the instruction is executed in an electronic device, the program causes the electronic device to execute the operation steps included in the method of the present invention.

Embodiments of the present invention further provide a storage medium storing computer-readable instructions, where the computer-readable instructions cause an electronic device to execute the operation steps included in the method of the present invention.

It should be understood that, in various embodiments of the present invention, the sequence numbers of the above-mentioned processes do not mean the execution sequence, and the execution sequence of each process should be determined by its function and inherent logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

Those of ordinary skill in the art will appreciate that the elements and algorithm steps of the examples described in connection with the embodiments disclosed herein may be embodied in electronic hardware, computer software, or combinations of both, and that the components and steps of the examples have been described in a functional general in the foregoing description for the purpose of illustrating clearly the interchangeability of hardware and software. Whether such functionality is implemented as hardware or software depends upon the particular application and design constraints imposed on the implementation. Skilled artisans may implement the described functionality in varying ways for each particular application, but such implementation decisions should not be interpreted as causing a departure from the scope of the present invention.

It is clear to those skilled in the art that, for convenience and brevity of description, the specific working processes of the above-described systems, apparatuses and units may refer to the corresponding processes in the foregoing method embodiments, and are not described herein again.

In the several embodiments provided in the present application, it should be understood that the disclosed system, apparatus and method may be implemented in other ways. For example, the above-described apparatus embodiments are merely illustrative, and for example, the division of the units is only one logical division, and other divisions may be realized in practice, for example, a plurality of units or components may be combined or integrated into another system, or some features may be omitted, or not executed. In addition, the shown or discussed mutual coupling or direct coupling or communication connection may be an indirect coupling or communication connection through some interfaces, devices or units, and may also be an electric, mechanical or other form of connection.

The units described as separate parts may or may not be physically separate, and parts displayed as units may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of the embodiment of the present invention.

The integrated unit, if implemented in the form of a software functional unit and sold or used as a stand-alone product, may be stored in a computer readable storage medium. Based on such understanding, the technical solution of the present invention essentially or partially contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product stored in a storage medium and including instructions for causing a computer device (which may be a personal computer, a server, or a network device) to execute all or part of the steps of the method according to the embodiments of the present invention. And the aforementioned storage medium includes: a U-disk, a removable hard disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), a magnetic disk or an optical disk, and other various media capable of storing program codes.

The above description is only for the specific embodiments of the present invention, but the scope of the present invention is not limited thereto, and any person skilled in the art can easily conceive of the changes or substitutions within the technical scope of the present invention, and all the changes or substitutions should be covered within the scope of the present invention.

Claims

1. A system for verifying radiation therapy dosage, comprising:

2. The system of claim 1, wherein the feature screening module comprises:

3. The system of claim 1, wherein the radiology model is trained by: