MLC QA Integration Method Based on Fence Testing

Abstract

This invention relates to the field of quality control technology for medical linear accelerators, specifically an integrated MLC QA method based on fence testing. The method includes: acquiring fence test images; establishing a beam radiation scattering library to store the radiation scattering amount corresponding to each MLC leaf pair in each strip; selecting the corresponding radiation scattering amount from the scattering library based on the strip width and spacing, and subtracting it from the image pixel values ​​to partially eliminate inter-strip beam radiation scattering; extracting algorithm parameters from the pixel value profile curve after radiation scattering elimination; and substituting the algorithm parameters into a pre-generated fitting function to calculate the actual gap or physical position of the MLC leaf. This invention partially eliminates inter-strip radiation scattering interference by establishing a beam radiation scattering library, and combines multiple algorithm parameters for leaf position detection, improving the accuracy and robustness of MLC quality assurance. It also solves the problem that existing half-peak height-width ratio (WHM) methods are susceptible to pixel value instability and inter-strip beam radiation scattering.

Description

Technical Field

[0001] This invention belongs to the field of quality control technology for medical linear accelerators, and particularly relates to an MLCQA integration method based on fence testing. Background Technology

[0002] Radiotherapy is one of the main treatments for malignant tumors. Medical linear accelerators are the core equipment for radiotherapy, and their integrated multi-leaf collimator (MLC) precisely shapes the beam, ensuring a high-dose region is highly conformal to the tumor target area while maximizing the protection of surrounding organs at risk. Therefore, the positional accuracy of the MLC blades directly determines the accuracy and safety of radiotherapy. To ensure the positioning performance of the MLC, regular quality assurance (QA) testing is required.

[0003] Picket fence testing is a widely used method in MLC QA. This test analyzes the positional consistency of blade edges at the boundaries of multiple parallel stripes by generating a field image. Picket fence testing is typically performed on film or an electronic portal imaging device (EPID). While film offers high spatial resolution and is considered the gold standard for MLC QA, its operation is cumbersome, prone to human error, and costly in terms of time and money. In contrast, EPID can directly acquire digital images and is integrated with the accelerator system without requiring additional setup, thus its application in MLC QA is becoming increasingly popular.

[0004] In existing technologies, the FWHM method is commonly used for analyzing fence test images. This method extracts the pixel value profile curve of each strip in the image, calculates the width between the half-peaks on both sides of the curve, and then infers the blade gap and blade position. However, the FWHM method has significant limitations in practical applications. First, this method is highly dependent on the stability of pixel values, while EPD images are susceptible to interference from noise, dead pixels, and other factors. The film handling process may also introduce errors, leading to a decrease in the accuracy of blade edge detection or even failure. In addition, existing research shows that there is beam radiation scattering between strips in fence test images. This scattering interferes with the pixel value distribution of the target strips, causing the analysis results to be related to the position of the strips in the image, thus affecting the accuracy and consistency of the measurement results. Currently, most commercial MLC QA software is a closed system with its algorithms not publicly available. Users cannot personalize the test mode and algorithm according to actual needs, and there is a lack of technical solutions to effectively eliminate interference from inter-strip radiation scattering.

[0005] In summary, existing fence testing techniques based on the FWHM method face problems such as unstable pixel values, poor anti-interference ability, inability to shield inter-strip radiation scattering, and poor algorithm customizability in MLC QA. There is an urgent need to develop an improved MLC QA method that is robust, accurate, and can partially eliminate scattering interference. Summary of the Invention

[0006] To address the shortcomings of existing technologies, the purpose of this invention is to provide an integrated MLC QA method based on fence testing, thereby solving the problems faced by existing fence testing technologies in MLC QA, such as unstable pixel values, poor anti-interference ability, inability to shield inter-strip radiation scattering, and poor algorithm customizability.

[0007] To solve the above-mentioned technical problems, the present invention adopts the following technical solution:

[0008] This invention provides an MLC QA integration method based on fence testing, comprising the following steps:

[0009] S10. Obtain a fence test image, wherein the fence test image contains multiple beam stripes;

[0010] S20. Establish a beam radiation scattering library, which stores the radiation scattering amount corresponding to each MLC leaf pair in each strip. The radiation scattering amount is determined based on the pixel value difference of the EPD image obtained in the independent delivery mode and the composite delivery mode.

[0011] S30. For the fence test image to be analyzed, select the corresponding radiation scattering amount from the beam radiation scattering library according to the width and spacing of the strips, and subtract the radiation scattering amount from the image pixel value to partially eliminate beam radiation scattering between strips.

[0012] S40. On the pixel value profile curve after eliminating radiation scattering, extract at least one algorithm parameter, the algorithm parameter including at least one of peak height, full-width half-maximum (FWHM) and area under the peak;

[0013] S50. Substitute the extracted algorithm parameters into the pre-generated fitting function to calculate the actual gap or physical position of the MLC blade.

[0014] Furthermore, in S20, the specific steps for establishing the beam radiation scattering library are as follows:

[0015] S201. Design a set of QA plans that includes multiple fence test modes, each test mode containing multiple QA plans with different strip widths and strip spacings;

[0016] S202. Deliver each QA plan separately using independent delivery mode and composite delivery mode, and obtain the corresponding EPID image;

[0017] S203. Subtract the EPD image pixel values ​​obtained in the composite delivery mode from the EPD image pixel values ​​obtained in the independent delivery mode to obtain the radiation scattering amount of each MLC leaf pair in each strip.

[0018] S204. Store the radiation scattering quantities according to the strip number and MLC leaf pair number to form a beam radiation scattering library.

[0019] Furthermore, in the QA program set, each QA program contains 7 beam strips, with a strip width of 7mm to 13mm, a strip length increase gradient of 1mm, and a strip spacing of not less than 16mm.

[0020] Furthermore, before extracting the algorithm parameters, a step of normalizing the EPD image is included: the acquired EPD image is normalized using an open field image of size 28cm×28cm to eliminate variations in the daily output of the medical linear accelerator and the flatness and symmetry of the beam.

[0021] Furthermore, before extracting the algorithm parameters, a preprocessing step for the pixel value profile curve is included: the average value of the middle 5 rows of EPD pixel values ​​corresponding to MLC is taken as the pixel value profile curve for subsequent analysis.

[0022] Furthermore, when extracting the area under the peak as an algorithm parameter, the specific steps include: subtracting the pixel values ​​on both sides of the peak of the curve from the minimum pixel value on the same side, and summing the differences to obtain the area under the curve.

[0023] Furthermore, when extracting algorithm parameters, only a limited number of EPD pixel values ​​are selected for analysis. Unselected pixels are not included in subsequent data analysis. The number of selected pixels is determined by the following formula:

[0024]

[0025] Where C is the calculated number of pixels, m is the distance from the radiation source to the detector that is proportional to the distance from the source to the isocenter, g is the width of the strip MLC leaf pair gap set in the Treatment Planning System (TPS), s is the physical spacing between pixels on the EPD board, and i and j are the strip number and MLC leaf pair number in the fence test image, respectively.

[0026] Furthermore, the extracted algorithm parameters are fitted with a fourth-order function to the actual MLC leaf pair gaps extracted from the log file. The fitting formula is as follows:

[0027]

[0028] in, denoted as the coefficients of the polynomial, and x is the parameter of each algorithm.

[0029] Furthermore, it also includes the step of fusing multiple basic algorithms using ensemble learning methods, specifically including:

[0030] Voting integration method: Randomly select a portion of the results from multiple error QA plan measurements, sort them to select the basic algorithm with the best position accuracy and store it. In subsequent measurement and analysis, the stored basic algorithm is used for MLC blade position detection.

[0031] Fitting ensemble method: Select the peak area method and FWHM method for multiple measurement results and perform a fourth-order function fitting, retain the fitting parameters, and use the function to detect the position of MLC blades in subsequent analysis.

[0032] Furthermore, during the generation of the fitting function and the algorithm verification process, the actual position of the MLC blade is based on the blade position recorded in the medical linear accelerator log file.

[0033] The MLC QA integration method based on fence testing provided by this invention has at least the following advantages compared with the prior art:

[0034] Existing fence testing techniques in MLC QA suffer from problems such as unstable pixel values, poor anti-interference capabilities, inability to shield inter-strip radiation scattering, and poor algorithm customizability. This invention integrates multiple basic algorithms, improving algorithm accuracy and robustness. It solves the problem of FWHM method being prone to MLC QA test failure due to unstable pixel values, thus improving upon FWHM. Furthermore, based on an established scattering library, this invention partially eliminates inter-strip beam radiation scattering in fence test images, addressing the issue that the accuracy of fence test image analysis results is related to strip position. During algorithm development, log files are used to assist in algorithm development and verification, eliminating the impact of random blade errors on algorithm development, improving algorithm accuracy, and ensuring the scientific rigor and precision of the entire algorithm development and verification process. Attached Figure Description

[0035] To more clearly illustrate the solution of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. The drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0036] Figure 1A flowchart of the MLC QA integration method based on fence testing provided in an embodiment of the present invention;

[0037] Figure 2 The beam strip design diagram of each QA plan in TPS for the MLC QA integration method based on fence testing provided in the embodiments of the present invention;

[0038] Figure 3 The EPID image of each beam strip in the independent delivery mode of the MLC QA integration method based on fence testing provided in the embodiments of the present invention;

[0039] Figure 4 The EPID image of each beam strip in the composite delivery mode of the MLC QA integration method based on fence testing provided in the embodiments of the present invention;

[0040] Figure 5 A normalized EPD row pixel value profile curve in the MLC QA integration method based on fence testing provided in this embodiment of the invention;

[0041] Figure 6 A graph showing the selection of a limited number of EPD pixel values ​​for analysis in the MLC QA integration method based on fence testing provided in this embodiment of the invention.

[0042] Figure 7 This is a schematic diagram illustrating the algorithm development principle in the beam adjacency region of the fence test image in the MLC QA integration method based on fence testing provided in this embodiment of the invention. Detailed Implementation

[0043] To facilitate understanding of the present invention, a more complete description will be given below with reference to the accompanying drawings. Preferred embodiments of the invention are shown in the drawings. However, the invention can be implemented in many different forms and is not limited to the embodiments described herein. Rather, these embodiments are provided to provide a thorough and complete understanding of the disclosure of the invention.

[0044] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. The terminology used herein in the description of the invention is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention.

[0045] This invention provides an MLC QA integration method based on barrier testing, applied to the MLC QA inspection process. The MLC QA integration method based on barrier testing includes the following steps:

[0046] S10. Acquire a fence test image containing multiple beam strips; S20. Establish a beam radiation scattering library, which stores the radiation scattering amount corresponding to each MLC blade pair in each strip. The radiation scattering amount is determined based on the pixel value difference between the EPD images acquired in independent delivery mode and composite delivery mode; S30. For the fence test image to be analyzed, select the corresponding radiation scattering amount from the beam radiation scattering library according to the width and spacing of the strips, and subtract the radiation scattering amount from the image pixel value to partially eliminate beam radiation scattering between strips; S40. Extract at least one algorithm parameter from the pixel value profile curve after eliminating radiation scattering. The algorithm parameter includes at least one of peak height, FWHM, and area under the peak; S50. Substitute the extracted algorithm parameter into a pre-generated fitting function to calculate the actual gap or physical position of the MLC blades.

[0047] This invention partially eliminates inter-strip radiation scattering interference by establishing a beam radiation scattering library, and improves the accuracy and robustness of MLC QA by combining multiple algorithm parameters for blade position detection. It also solves the problem that the existing FWHM method is susceptible to unstable pixel values ​​and the influence of inter-strip beam radiation scattering.

[0048] To enable those skilled in the art to better understand the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings.

[0049] This invention provides an integrated method for MLC QA based on fence testing, applied in the MLC QA detection process, combining... Figures 1 to 6 In this embodiment, the MLC QA integration method based on fence testing includes the following steps:

[0050] S10. Obtain a fence test image, wherein the fence test image contains multiple beam strips.

[0051] Specifically, in this embodiment, a series of fence test pattern sets are designed in the TPS. Each test pattern contains 7 QA plans, each plan contains 7 beam strips (beam irradiation areas) with a width of 24cm, and each QA plan contains only one strip width and strip spacing, such as... Figure 2 As shown, the strip length is 7mm to 13mm, with a strip length increase gradient of 1mm. That is, each fence test mode contains a total of 7 QA plans with 7 beam strips, and the spacing between the beam strips is not less than 16mm.

[0052] Each QA program is delivered via the medical linear accelerator using two delivery modes: independent delivery (…). Figure 3 ) and composite delivery ( Figure 4In standalone delivery mode, EPD acquires only one beamstrip, and the image data obtained comes solely from a single irradiation by the medical accelerator. Therefore, it does not include beam radiation scattering from other beamstrip deliveries. In composite delivery mode, a single EPD image contains seven beams, and the image data for each beams includes beam radiation scattering from other beamstrip deliveries. The QA program will execute weekly for a total of nine times over three months in both delivery modes, and will save the relevant data.

[0053] S20. Establish a beam radiation scattering library, which stores the radiation scattering amount corresponding to each MLC leaf pair in each strip. The radiation scattering amount is determined based on the pixel value difference of the EPD image obtained in independent delivery mode and composite delivery mode.

[0054] Specifically, in this embodiment, step S20 consists of the following steps:

[0055] S201. Design a set of QA plans that includes multiple fence test modes, each test mode containing multiple QA plans with different strip widths and strip spacings;

[0056] S202. Deliver each QA plan separately using independent delivery mode and composite delivery mode, and obtain the corresponding EPID image;

[0057] S203. Subtract the EPD image pixel values ​​obtained in the composite delivery mode from the EPD image pixel values ​​obtained in the independent delivery mode to obtain the radiation scattering amount of each MLC leaf pair in each strip.

[0058] S204. Store the radiation scattering quantities according to the strip number and MLC leaf pair number to form a beam radiation scattering library.

[0059] In this embodiment, there is no uniform standard for the design of beam strip width and spacing. Narrower strip width or larger strip spacing is applicable to both adjacent areas and irradiation areas in this embodiment of the invention.

[0060] In this embodiment, to eliminate variations in the daily output of the medical linear accelerator and the flatness and symmetry of the beam, the obtained EPID image was normalized with an open field of 28cm × 28cm. The normalized EPID row pixel value profile curve is shown below. Figure 5 As shown. To eliminate the influence of beam leakage between MLC blades on the EPD pixel values, the average of the middle 5 rows of EPD pixel values ​​corresponding to the MLC was taken and used as the basis for subsequent analysis. The difference in EPD pixel values ​​between the two delivery modes is taken as the beam radiation scattering in that mode, as shown. Figure 5As shown, the radiative scattering of each strip and each MLC leaf pair is stored separately, forming a radiative scattering library.

[0061] S30. For the fence test image to be analyzed, select the corresponding radiation scattering amount from the beam radiation scattering library according to the width and spacing of the strips, and subtract the radiation scattering amount from the image pixel value to partially eliminate beam radiation scattering between strips.

[0062] S40. On the pixel value profile curve after eliminating radiation scattering, extract at least one algorithm parameter, the algorithm parameter including at least one of peak height, FWHM and area under the peak.

[0063] S50. Substitute the extracted algorithm parameters into the pre-generated fitting function to calculate the actual gap or physical position of the MLC blade.

[0064] Specifically, in this embodiment, to further reduce the impact of beam radiation scattering on the development of the new algorithm, only a limited number of EPID pixel values ​​are selected for analysis in each MLC leaf pair analysis region of the normalized pixel value profile curve. Unselected pixels will no longer participate in subsequent data analysis. Figure 6 As shown. The selected pixel data follows the following formula:

[0065]

[0066] Where C represents the calculated number of pixels, m is the distance from the radiation source to the detector proportional to the distance from the source to the isocenter, g represents the width of the strip MLC leaf pair gap set in TPS, s represents the physical spacing between pixels on the EPID board, and i and j represent the strip number and MLC leaf pair number in the fence test image, respectively. On the normalized pixel value curve with a finite number of pixels, the FWHM of the FWHM method is still the distance between the left and right half-peak heights; to increase the sensitivity of the algorithm, only the peak value is selected as the algorithm parameter for modeling; to increase the robustness of the algorithm, the area covered by the curve is used as the algorithm parameter; to reduce the complexity of the algorithm, the pixel values ​​on both sides of the peak value of the curve are subtracted from the minimum pixel value on the same side, and the sum is used as the area under the curve. The development principle of each algorithm parameter is as follows: Figure 6 As shown. The algorithm parameters are fitted with a fourth-order function to the actual MLC leaf pair gap, and the fitting formula is as follows:

[0067]

[0068] in, The coefficients of the polynomial are represented by , and x represents each parameter of the algorithm.

[0069] Because the blade position is not always the same as the preset position in the TPS during irradiation in a medical linear accelerator, the blade position obtained from analysis of the log files is used as the standard throughout the algorithm development and verification process.

[0070] Furthermore, in this embodiment, to test the accuracy of each algorithm, a leaf pair position error sequence is introduced along the direction of leaf movement in a fence test image with a strip width of 10mm. This error sequence includes leaf pair position errors of 0.4mm, 0.8mm, 1.2mm, 1.6mm, 1.8mm, 2.4mm, and 2.8mm. Each leaf pair position error is a set of multiple leaf position errors. For example, a 0.8mm leaf pair error includes combinations of leaf position errors such as (-0.8, 0), (0, -0.8), (-0.4, 0.4), and (0.4, 0.4), where a negative sign indicates under-travel and a positive sign indicates over-travel. The error QA plan is executed only once a week in the composite delivery mode, for a total of 9 times.

[0071] Before using the algorithms mentioned above to detect the position of MLC blades in the EPID images from the fence test based on the error QA program, it is necessary to select the corresponding scattering amount from the scattering library according to the width and spacing of the stripes, and subtract this scattering amount from the MLC blade pairs. Then, the edges of the MLC blades are detected according to the three basic algorithms mentioned above. To further enhance the accuracy and robustness of the algorithms, an ensemble learning approach is adopted. The measurement results of nine error QA programs from the three basic algorithms are integrated using voting and function fitting methods, respectively.

[0072] Voting integration method: From the 9 error QA plan measurement results, 5 results are randomly selected, and the basic algorithm with the best position accuracy is selected by sorting and stored as "memory". In subsequent measurement analysis, the counting algorithm of this memory is used to detect the position of MLC blades.

[0073] The fitting ensemble method was employed. Due to the high sensitivity of the peak height method, it exhibits higher uncertainty compared to the other two basic algorithms. Therefore, from the nine error QA plan measurement results, five results from the peak area and FWHM methods were randomly selected for four-fold function fitting, retaining relevant parameters. This function was used for MLC blade position detection in subsequent analyses.

[0074] In some other embodiments, such as Figure 7 As shown, the MLC QA fence test can also analyze the critical zone between shooting fields, and the size of the analysis area can be controlled within 5cm. The related experimental design, testing and algorithm development process is completely consistent with that in this embodiment.

[0075] The MLC QA integration method based on fence testing described in the above embodiments addresses the shortcomings of existing technologies in MLC QA, such as unstable pixel values, poor anti-interference capabilities, inability to shield inter-strip radiation scattering, and poor algorithm customizability. This invention integrates multiple basic algorithms, improving algorithm accuracy and robustness, and solving the problem of MLC QA test failure due to unstable pixel values ​​in the FWHM method, thus improving upon the FWHM method. Furthermore, based on an established scattering library, this invention partially eliminates inter-strip beam radiation scattering in fence test images, resolving the issue that the accuracy of fence test image analysis results is related to strip position. During algorithm development, log files are used to assist in algorithm development and verification, eliminating the impact of random blade errors on algorithm development, improving algorithm accuracy, and ensuring the scientific rigor and comprehensiveness of the entire algorithm development and verification process.

[0076] Obviously, the embodiments described above are merely preferred embodiments of the present invention, and not all embodiments. The accompanying drawings illustrate preferred embodiments of the present invention, but do not limit the scope of the patent. The present invention can be implemented in many different forms; rather, these embodiments are provided to provide a more thorough and complete understanding of the disclosure of the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing specific embodiments, or make equivalent substitutions for some of the technical features. Any equivalent structures made using the content of this specification and drawings, directly or indirectly applied to other related technical fields, are similarly within the scope of patent protection of this invention.

Claims

1. An MLC QA integration method based on fence testing, characterized in that, Includes the following steps: S10. Obtain a fence test image, wherein the fence test image contains multiple beam stripes; S20. Establish a beam radiation scattering library, which stores the radiation scattering amount corresponding to each MLC leaf pair in each strip. The radiation scattering amount is determined based on the pixel value difference of the EPD image obtained in the independent delivery mode and the composite delivery mode. S30. For the fence test image to be analyzed, select the corresponding radiation scattering amount from the beam radiation scattering library according to the width and spacing of the strips, and subtract the radiation scattering amount from the image pixel value to partially eliminate beam radiation scattering between strips. S40. On the pixel value profile curve after eliminating radiation scattering, extract at least one algorithm parameter, the algorithm parameter including at least one of peak height, half-peak width and height and area under the peak; S50. Substitute the extracted algorithm parameters into the pre-generated fitting function to calculate the actual gap or physical position of the MLC blade.

2. The MLC QA integration method based on fence testing according to claim 1, characterized in that, In step S20, the specific steps for establishing the beam radiation scattering library are as follows: S201. Design a set of QA plans that includes multiple fence test modes, each test mode containing multiple QA plans with different strip widths and strip spacings; S202. Deliver each QA plan separately using independent delivery mode and composite delivery mode, and obtain the corresponding EPID image; S203. Subtract the EPD image pixel values ​​obtained in the composite delivery mode from the EPD image pixel values ​​obtained in the independent delivery mode to obtain the radiation scattering amount of each MLC leaf pair in each strip. S204. Store the radiation scattering quantities according to the strip number and MLC leaf pair number to form a beam radiation scattering library.

3. The MLC QA integration method based on fence testing according to claim 2, characterized in that, In the QA program set, each QA program contains 7 beam strips with a strip width of 7mm to 13mm, a strip length increase gradient of 1mm, and a strip spacing of not less than 16mm.

4. The MLC QA integration method based on fence testing according to claim 1, characterized in that, Before extracting the algorithm parameters, the process also includes a step of normalizing the EPID image: the acquired EPID image is normalized using an open field image of size 28cm×28cm to eliminate variations in the daily output of the medical linear accelerator and the flatness and symmetry of the beam.

5. The MLC QA integration method based on fence testing according to claim 1, characterized in that, Before extracting the algorithm parameters, a preprocessing step for the pixel value profile curve is also included: the average value of the middle 5 rows of EPID pixel values ​​corresponding to MLC is taken as the pixel value profile curve for subsequent analysis.

6. The MLC QA integration method based on fence testing according to claim 1, characterized in that, When extracting the area under the peak as an algorithm parameter, the specific steps include: subtracting the pixel values ​​on both sides of the peak value of the curve from the minimum pixel value on the same side, and summing the differences to obtain the area under the curve.

7. The MLC QA integration method based on fence testing according to claim 1, characterized in that, When extracting algorithm parameters, only a limited number of EPD pixel values ​​are selected for analysis. Unselected pixels are not included in subsequent data analysis. The number of selected pixels is determined by the following formula: Where C is the calculated number of pixels, m is the distance from the radiation source to the detector that is proportional to the distance from the source to the isocenter, g is the width of the strip MLC leaf pair gap set in the treatment planning system, s is the physical spacing between pixels on the EPD board, and i and j are the numbers of the strip and the MLC leaf pair in the fence test image, respectively.

8. The MLC QA integration method based on fence testing according to claim 1, characterized in that, The extracted algorithm parameters are fitted with a fourth-order function to the actual MLC leaf pair gaps extracted from the log file. The fitting formula is as follows: in, denoted as the coefficients of the polynomial, and x is the parameter of each algorithm.

9. The MLC QA integration method based on fence testing according to claim 1, characterized in that, It also includes the step of fusing multiple basic algorithms using ensemble learning methods, specifically including: Voting integration method: Randomly select a portion of the results from multiple error QA plan measurements, sort them to select the basic algorithm with the best position accuracy and store it. In subsequent measurement and analysis, the stored basic algorithm is used for MLC blade position detection. Fitting ensemble method: Select the peak area method and FWHM method for multiple measurement results and perform a fourth-order function fitting, retain the fitting parameters, and use the function to detect the position of MLC blades in subsequent analysis.

10. The MLC QA integration method based on fence testing according to claim 1, characterized in that, During the generation of the fitting function and algorithm verification, the actual position of the MLC blade is based on the blade position recorded in the medical linear accelerator log file.

Images