Radiation dose adjustment system based on deep learning

Abstract

The present invention relates to the field of neural network technology, and specifically to a radiation dose adjustment system based on deep learning. The system includes: an image analysis module that scans and distinguishes the boundaries between tumors and normal tissues based on medical image data input, refines the size and shape of the tumor in the image through edge detection, extracts key features in the image through contrast enhancement, and generates tumor feature data. In the present invention, by distinguishing the boundaries between tumors and normal tissues and refining the features of the tumor in the image, the distribution of radiation doses is made more accurate and personalized. This detailed image analysis optimizes the treatment plan, ensures that the dose is concentrated in the lesion area, and reduces the impact on healthy tissues. In addition, the analysis of integrated clinical parameters enhances the understanding of the patient's radiation sensitivity and provides a treatment plan that better meets individual needs. This dose adjustment method based on in-depth analysis effectively adjusts the dose deviation in treatment by real-time monitoring of the execution effect of radiotherapy.

Description

Technical Field

[0001] The present invention relates to the field of neural network technology, and in particular to a radiation dose adjustment system based on deep learning. Background Art

[0002] Neural networks are computational models inspired by biological brains that simulate the processing methods of the human brain to perform various machine learning tasks. These networks consist of a large number of nodes, or neurons, which are connected in a hierarchical structure and can learn to recognize patterns and features through training. Neural network technology has been widely used in many fields, including speech recognition, image processing, medical diagnosis, etc. With the improvement of hardware performance and the development of algorithms, neural network models, especially deep learning models, have become one of the key technologies to promote the frontier of artificial intelligence.

[0003] Among them, the radiation dose adjustment system refers to the use of artificial intelligence technologies such as neural networks to optimize the dose distribution in radiotherapy. The purpose of this type of system is to ensure that patients receive appropriate and accurate radiation doses in the disease-affected area during radiotherapy, while minimizing radiation exposure to surrounding healthy tissues. This technology can improve treatment effects and reduce side effects, and is an important development direction in the field of modern medical radiation.

[0004] Existing technologies often lack sufficient accuracy in radiotherapy dose distribution and rely mainly on experience and standardized dose plans, which may lead to inaccurate irradiation of tumors and unnecessary radiation exposure to surrounding healthy tissues. For example, failure to accurately distinguish the boundaries between tumors and normal tissues may lead to insufficient or excessive dose coverage, affecting the treatment effect and increasing side effects. In addition, existing technologies lack an in-depth understanding of individual differences and feedback mechanisms, and are unable to adjust dose deviations during treatment in real time, which may make it difficult for treatment plans to adapt to the actual conditions of patients during implementation. These limitations reduce the efficiency and safety of treatment to a certain extent and increase the treatment burden on patients. Summary of the invention

[0005] The purpose of the present invention is to solve the shortcomings existing in the prior art and propose a radiation dose adjustment system based on deep learning.

[0006] In order to achieve the above-mentioned purpose, the present invention adopts the following technical solution: a radiation dose adjustment system based on deep learning includes:

[0007] The image analysis module scans and distinguishes the boundary between tumors and normal tissues based on medical image data input, refines the size and shape of tumors in images through edge detection, extracts key features in images through contrast enhancement, and generates tumor feature data;

[0008] The clinical parameter analysis module is based on the tumor characteristic data, collects the patient's body position information, tumor type and previous radiation reaction records, analyzes and calculates the tumor's radiosensitivity index, and obtains the clinical parameter coordination result;

[0009] The dose decision module is based on the clinical parameter synergy results and compared with the existing radiation dose database. By analyzing the correlation between pathological characteristics and clinical data, it adjusts the individualized dose distribution, refines the matching degree and adaptability of the dose, and obtains the dose adjustment plan;

[0010] The dose implementation monitoring module monitors the dose implementation effect during the radiotherapy process in real time based on the dose adjustment scheme, records the dynamic data of the treatment, adjusts the dose deviation during the treatment, and generates an implementation effect report.

[0011] As a further solution of the present invention, the step of acquiring the tumor characteristic data is specifically as follows:

[0012] For the input medical image data, Gaussian smoothing is performed using the formula:

[0013] ;

[0014] Generate smoothed image ;

[0015] in, Represents the smoothed image, which is obtained by inputting the medical image Using Gaussian kernel After convolution, represents the Gaussian kernel, which is used to smooth the image data to reduce noise. represents the input medical image;

[0016] Based on the smoothed image, the Canny algorithm is used for edge detection, using the formula:

[0017] ;

[0018] Generate edge enhancement map ;

[0019] in, Represents the edge enhancement map, through use Operator and Calculate separately and The gradient of the direction is synthesized into an edge map. Representatives in Directional Operator, used to detect Direction Edge Representatives in Directional Operator, used to detect Direction edge;

[0020] The edge enhancement image is enhanced using a contrast enhancement algorithm, using the formula:

[0021] ;

[0022] Generate contrast-enhanced image ;

[0023] in, Represents the contrast enhancement map, through Perform normalization and contrast adjustment to enhance image contrast and Respectively represent The maximum and minimum values ​​of Represents the contrast adjustment coefficient, which is used to adjust the degree of contrast enhancement;

[0024] Extract and summarize key features from the contrast-enhanced image, combined with the size and morphology information of the tumor, using the formula:

[0025] ;

[0026] Generating tumor characterization data ;

[0027] in, Represents tumor characteristic data, through The features in are weighted summed to obtain the final feature value. Representative The weight of a feature indicates the importance of the feature in tumor detection and analysis. Represents the weighted reinforcement index, which is used to adjust the contribution of the feature. Contrast enhancement image The Features.

[0028] As a further solution of the present invention, the steps for obtaining the clinical parameter synergy results are specifically as follows:

[0029] The tumor characteristic data and the patient's body position information are integrated using the formula:

[0030] ;

[0031] Computational generation of the initial synthetic dataset ;

[0032] in, represents the initial comprehensive dataset, which combines tumor characteristic data and patient position information to form, Represents the adjustment coefficient, used to adjust exist Contribution in Represents the adjustment coefficient, which is used to adjust the contribution of the cosine value of the body position information. Position information The cosine value calculated after conversion to radians affects the sensitivity of the data set. Represents the adjustment coefficient, used to adjust The square value of Contribution to the results;

[0033] The initial comprehensive data were combined with tumor type using the formula:

[0034] ;

[0035] Generate intermediate dataset ;

[0036] in, Represents the intermediate data set, which is obtained by Processing and tumor type The association process is used to generate Represents the weight coefficient, used to adjust The squared contribution, Represents the weight coefficient, used to adjust The square root of Impact on results;

[0037] With reference to the intermediate data set and the previous radiation response records, the formula is used:

[0038] ;

[0039] Radiosensitivity index ;

[0040] in, Represents the radiosensitivity index, which is integrated and previous radiation response records To evaluate, Represents the coefficient for adjusting the sensitivity of radiation response, affecting The decay rate of the radioactive response is adjusted to Represents the tantalum sensitivity threshold, used to adjust the sensitivity function The offset in .

[0041] As a further solution of the present invention, the step of obtaining the correlation between the pathological characteristics and the clinical data is specifically:

[0042] Based on the clinical parameters synergistic results and the existing radiation dose database, the formula was used:

[0043] ;

[0044] Calculate the initial relevance score ;

[0045] in, stands for initial correlation score, which is calculated by calculating the radiosensitivity index The square of the radiation dose The product of , and then multiplied by the weight factor of each sample , then sum and divide by the number of samples Got it, Represents the radiation dose value and is used to quantify the dose level of radiotherapy. Represents the weight factor of each sample, which is used to adjust the influence of each sample in the calculation. Represents the sample size and is used to calculate the mean;

[0046] The initial relevance score is iteratively analyzed using the formula:

[0047] ;

[0048] Perform normalization to fit the different data ranges and generate standardized scores ;

[0049] in, represents the standardized score, which is obtained by scoring the initial relevance A linear transformation is performed to obtain Minimum value of and maximum value Normalize by adding a small constant To avoid division by zero, and Respectively represent The minimum and maximum values ​​of range, represents a small constant used to avoid a zero denominator;

[0050] Using the standardized scores and the variance of the clinical dataset, the formula was used:

[0051] ;

[0052] Calculate the Pearson correlation coefficient to determine the correlation between pathological features and clinical data ;

[0053] in, stands for Pearson correlation coefficient, which is calculated by normalizing the scores and clinical data The sum of the products of the deviations and then divided by and The square root of the sum of the squares of the deviations is obtained, where Represents clinical data and is used to measure the correlation with pathological features. and Respectively represent and The mean of is used to calculate the deviation.

[0054] As a further solution of the present invention, the step of obtaining the dosage adjustment plan is specifically:

[0055] Based on the correlation between the pathological features and clinical data, the initial dose adjustment was performed using the formula:

[0056] ;

[0057] Calculate preliminary dose adjustments ;

[0058] in, Represents the initial dose adjustment value, which is based on the basic dose and Pearson correlation coefficient Calculated represents the patient's basic radiation dose, Represents the adjustment coefficient, which is used to adjust Calculated correction factor The intensity of the impact;

[0059] The preliminary dose adjustment value is refined with reference to individual differences among patients, using the formula:

[0060] ;

[0061] Adjust the dose to match individual response and get dose adjustment plan ;

[0062] in, Represents the final dose adjustment value, which is refined To adapt to individual differences of patients, Represents the individual difference adjustment coefficient, which is used to adjust the intensity of patient side effects The square root of The impact on dose adjustments, thereby ensuring that dose adjustments more accurately reflect the patient's actual needs and responses.

[0063] As a further solution of the present invention, the steps of obtaining the implementation effect report are specifically as follows:

[0064] Based on the dose adjustment scheme, the dose implementation effect during radiotherapy is monitored in real time using the formula:

[0065] ;

[0066] Calculate and generate actual dose records ;

[0067] in, Represents the actual delivered dose record, which is based on the final dose adjustment value Calculated, where is the possible deviation ratio during implementation. and It is a newly added weight parameter used to adjust the actual dose according to real-time feedback to ensure accurate dose implementation;

[0068] Based on the actual dose record, the dynamic data during the treatment process is continuously recorded using the formula:

[0069] ;

[0070] Accumulate the implemented dose data and generate dynamic dose data ;

[0071] in, Represents the cumulative dose data, which is calculated by Weighted accumulation, the weight is determined by the duration of each time point and time weighting factor Decisions, which accurately reflect the dose accumulation throughout the treatment course;

[0072] Compare the dynamic dose data with the planned dose using the formula:

[0073] ;

[0074] Calculate the deviation ratio from the planned dose and generate implementation effect report ;

[0075] in, Represents the deviation ratio from the planned dose, which is calculated by calculating the cumulative implemented dose and planned dose and take the square root of it to obtain the standardized deviation ratio, where The planned dose is used to compare with the actual dose delivered to generate an implementation effect report .

[0076] Compared with the prior art, the advantages and positive effects of the present invention are:

[0077] In the present invention, by distinguishing the boundary between tumor and normal tissue and refining the characteristics of the tumor in the image, the distribution of radiation dose is made more accurate and personalized. This detailed image analysis optimizes the treatment plan, ensures that the dose is concentrated in the lesion area, and reduces the impact on healthy tissue. In addition, the analysis of integrated clinical parameters enhances the understanding of the patient's radiosensitivity and provides a treatment plan that better meets individual needs. This dose adjustment method based on in-depth analysis effectively adjusts the dose deviation in treatment by real-time monitoring of the execution effect of radiotherapy, thereby improving the adaptability and accuracy of treatment. BRIEF DESCRIPTION OF THE DRAWINGS

[0078] Figure 1 is a system flow chart of the present invention;

[0079] Figure 2 is a flow chart of the steps for obtaining tumor characteristic data of the present invention;

[0080] Figure 3 A flowchart of the steps for obtaining the clinical parameter synergy results of the present invention;

[0081] Figure 4 A flow chart of the steps for obtaining the correlation between pathological features and clinical data of the present invention;

[0082] Figure 5 A flow chart of the steps for obtaining the dosage adjustment scheme of the present invention;

[0083] Figure 6 The present invention is a flowchart of the steps for obtaining the implementation effect report of the present invention. DETAILED DESCRIPTION

[0084] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0085] In the description of the present invention, it should be understood that the terms "length", "width", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside" and the like indicate positions or positional relationships based on the positions or positional relationships shown in the drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as limiting the present invention. In addition, in the description of the present invention, "multiple" means two or more, unless otherwise clearly and specifically defined.

[0086] Embodiment 1:

[0087] See also Figure 1 , the radiation dose adjustment system based on deep learning includes:

[0088] The image analysis module scans and distinguishes the boundary between tumors and normal tissues based on medical image data input, refines the size and shape of tumors in images through edge detection, extracts key features in images through contrast enhancement, and generates tumor feature data;

[0089] The clinical parameter analysis module is based on tumor characteristic data, collects patient position information, tumor type and previous radiation reaction records, analyzes and calculates the tumor's radiosensitivity index, and obtains clinical parameter synergy results;

[0090] The dose decision module is based on the results of clinical parameter coordination and is compared with the existing radiation dose database. By analyzing the correlation between pathological characteristics and clinical data, it adjusts individualized dose allocation, refines the matching degree and adaptability of doses, and obtains a dose adjustment plan.

[0091] The dose implementation monitoring module monitors the dose implementation effect during radiotherapy in real time based on the dose adjustment plan, records the dynamic data of the treatment, adjusts the dose deviation during treatment, and generates an implementation effect report.

[0092] Tumor characteristic data specifically include tumor margins, morphological details, and key image features. Clinical parameter synergy results include body position information association, tumor type specificity, and early radiation response indicators. Dose adjustment plans include individualized dose allocation, dose matching, and dose adaptability. Implementation effect reports specifically include dose implementation effects, treatment dynamic data, and dose deviations during treatment.

[0093] See also Figure 2 ,The specific steps for obtaining tumor characteristic data are:

[0094] For the input medical image data, Gaussian smoothing is performed using the formula:

[0095] ;

[0096] Generate smoothed image ;

[0097] in, represents the Gaussian kernel, represents the input medical image;

[0098] Based on the smoothed image, the Canny algorithm is used for edge detection using the formula:

[0099] ;

[0100] Generate edge enhancement map ;

[0101] in, represents the Sobel operator in the x direction, represents the Sobel operator in the y direction;

[0102] Use contrast enhancement algorithm to enhance the edge enhancement image, using the formula:

[0103] ;

[0104] Generate contrast-enhanced image ;

[0105] in, represents the contrast adjustment factor;

[0106] Extract and summarize key features from the contrast-enhanced image, combined with the size and morphology information of the tumor, using the formula:

[0107] ;

[0108] Generating tumor characterization data ;

[0109] in, Representative The weight of the feature, represents the weighted reinforcement index, Contrast enhancement image The Features.

[0110] formula:

[0111] ;

[0112] Parameter details:

[0113] It is a Gaussian kernel used for image smoothing and is designed as a two-dimensional normal distribution function.

[0114] is the input medical image data.

[0115] Numerical example:

[0116] Set up a simplified one-dimensional image data , and the Gaussian kernel for .

[0117] Calculation process:

[0118] Convolution operation of Gaussian kernel and image data: The smoothed image data is calculated by adding the Gaussian kernel weights to the pixels around each pixel.

[0119] ;

[0120] This process does not perform calculations on edge pixels to avoid boundary effects, or can be handled by zero-filling the boundaries.

[0121] formula:

[0122] ;

[0123] Parameter details:

[0124] They are Sobel operators in the x and y directions, respectively, used to detect horizontal and vertical edges of images.

[0125] Numerical example:

[0126] set up and ,calculate edge detection.

[0127] Calculation process:

[0128] Horizontal edge detection: Yes application Operator.

[0129] Vertical edge detection: application Operator.

[0130] ;

[0131] Edge strength calculation:

[0132] ;

[0133] formula:

[0134] ;

[0135] Parameter details:

[0136] is the enhancement factor, which is used to adjust the intensity of contrast.

[0137] Numerical example:

[0138] set up ,and ,in and (Since there is only one point in the example).

[0139] Calculation process:

[0140] ;

[0141] formula:

[0142] ;

[0143] Parameter details:

[0144] are weight and reinforcement index respectively.

[0145] Numerical example:

[0146] set up ,and .

[0147] Calculation process:

[0148] ;

[0149] in conclusion:

[0150] The computational example shows how to generate tumor feature data from input image data through Gaussian smoothing, edge detection, and contrast enhancement. Represents a quantitative value of tumor feature data, with lower values ​​indicating that the feature is not significant.

[0151] See also Figure 3 , the specific steps for obtaining the clinical parameter collaborative results are:

[0152] Integrate tumor feature data with patient position information using the formula:

[0153] ;

[0154] Computational generation of the initial synthetic dataset ;

[0155] in, represents the adjustment coefficient, represents the adjustment coefficient, Represents the cosine value of the body position information converted into radians, represents the adjustment coefficient;

[0156] The initial comprehensive data were combined with tumor type using the formula:

[0157] ;

[0158] Generate intermediate dataset ;

[0159] in, represents the weight coefficient, represents the weight coefficient, square root treatment representing tumor type;

[0160] Referring to the intermediate data set and the previous radiation reaction records, the formula is used:

[0161] ;

[0162] Radiosensitivity index ;

[0163] in, represents the coefficient for adjusting the sensitivity of the radiation response, stands for radioactive reaction record, Represents the sensitivity threshold.

[0164] formula:

[0165] ;

[0166] in:

[0167] : Weight coefficient, which adjusts the influence of differentiated data sources on the results.

[0168] : Tumor characterization data that quantitatively describes the biomarker concentration or expression level of the tumor.

[0169] : Body position information, in degrees, describing the specific body position angle of the patient during treatment.

[0170] Set the value:

[0171] : Represents the quantitative value of tumor characteristic data.

[0172] Degrees: The patient's body position angle.

[0173] .

[0174] Calculation process:

[0175] ;

[0176] ;

[0177] ;

[0178] .

[0179] Result analysis:

[0180] : represents the initial data set after comprehensive reference to tumor characteristics and body position information, which is used for iterative analysis.

[0181] formula:

[0182] ;

[0183] in:

[0184] : weight coefficient;

[0185] : Tumor type, which represents differentiated tumor types in a numerical way.

[0186] Set the value:

[0187] : Set the tumor type to 4, which represents the target type or severity of the tumor;

[0188] .

[0189] Calculation process:

[0190] ;

[0191] ;

[0192] .

[0193] Result analysis:

[0194] : Represents the intermediate data set after referring to tumor characteristics, body position information and tumor type.

[0195] formula:

[0196] ;

[0197] in:

[0198] : sensitivity adjustment coefficient;

[0199] :Preliminary radiation response record, quantitative representation of biological response after radiotherapy;

[0200] : Threshold, used to adjust the sensitivity of the response;

[0201] Set the value:

[0202] : Quantitative value of the previous radioreaction record;

[0203] ;

[0204] Calculation process:

[0205] ;

[0206] ;

[0207] ;

[0208] Result analysis:

[0209] : Indicates the patient's radiation sensitivity index. The higher the value, the higher the patient's sensitivity to radiotherapy.

[0210] See also Figure 4 ,The steps for obtaining the correlation between pathological features and clinical data are as follows:

[0211] Based on the synergistic results of clinical parameters and the existing radiation dose database, the formula is used:

[0212] ;

[0213] Calculate the initial relevance score ;

[0214] in, Represents the radiation dose value, Represents the weight factor of each sample, represents the sample size;

[0215] The initial relevance score is iteratively analyzed using the formula:

[0216] ;

[0217] Perform normalization to fit the different data ranges and generate standardized scores ;

[0218] in, Respectively represent The minimum and maximum values ​​of represents a small constant used to avoid a zero denominator;

[0219] Using the standardized scores and the variance of the clinical dataset, the formula was used:

[0220] ;

[0221] Calculate the Pearson correlation coefficient to determine the correlation between pathological features and clinical data ;

[0222] in, represents the standardized score, Represents clinical data, Respectively represent and The mean of .

[0223] formula:

[0224] ;

[0225] Parameter Description:

[0226] : The square of the synergistic result of a patient's clinical parameters is set to the square of the concentration of a certain protein in the blood;

[0227] : The corresponding radiation dose in the database, such as the radiation dose used in treatment;

[0228] : Weight factor for each sample, set the weight to reflect the sample size or importance;

[0229] : Sample size, indicating the total number of data points involved in the calculation.

[0230] Derivation process and examples:

[0231] Set up data for three patients:

[0232] ;

[0233] ;

[0234] ;

[0235] calculate Values:

[0236] ;

[0237] Result analysis:

[0238] The obtained Represents the average relevance score after reference weights and is used in iterative analysis.

[0239] formula:

[0240] ;

[0241] Parameter Description:

[0242] : raw score;

[0243] :Fraction The minimum and maximum values ​​of

[0244] : A small constant to avoid zero denominator, set to 0.01.

[0245] Derivation process and examples:

[0246] Setting up on a larger sample set:

[0247] ;

[0248] calculate Values:

[0249] ;

[0250] Result analysis:

[0251] Represents the standardized score, used for display Position relative to the observation range, suitable for iterative correlation analysis.

[0252] formula:

[0253] ;

[0254] Parameter Description:

[0255] : Standardized scoring;

[0256] : Clinical datasets;

[0257] :They are and The mean of .

[0258] Derivation process and examples:

[0259] There are three sets of data points and the and The means of are 0.55 and 0.6 respectively. :

[0260] ;

[0261] (Set other data to balanced values ​​as well);

[0262] Result analysis:

[0263] here This means that no and This is because the example data is oversimplified or not representative enough.

[0264] See also Figure 5 , the specific steps for obtaining the dose adjustment plan are:

[0265] Based on the correlation between pathological features and clinical data, the initial dose adjustment is performed using the formula:

[0266] ;

[0267] Calculate preliminary dose adjustments ;

[0268] in, represents the basic dose, represents the adjustment coefficient;

[0269] Based on individual differences among patients, the preliminary dose adjustment value is refined using the formula:

[0270] ;

[0271] Adjust the dose to match individual response and get dose adjustment plan ;

[0272] in, represents the individual difference adjustment coefficient, The square root treatment represents the intensity of the patient's side effects.

[0273] formula:

[0274] ;

[0275] : Basic dose;

[0276] : Adjustment coefficient, used to adjust the sensitivity of the dose;

[0277] : Correlation analysis results.

[0278] Calculation process and examples:

[0279] Determine the base dose 50 Gy

[0280] Setting the adjustment factor 0.2

[0281] The correlation Cor obtained from the previous analysis was 0.8.

[0282] Substitute into the formula to calculate :

[0283] ;

[0284] The result, 50.9 Gy, represents the adjusted recommended treatment dose.

[0285] formula:

[0286] ;

[0287] : Obtained from the above results;

[0288] : Individual difference adjustment coefficient, set to 0.1;

[0289] : Intensity of side effects in patients.

[0290] Calculation process and examples:

[0291] The adjusted dose obtained from the above 50.9 Gy;

[0292] Setting the side effect strength is 4.

[0293] Substitute into the formula to calculate :

[0294] ;

[0295] The result is 61.08Gy, which represents the dose adjusted according to individual differences among patients.

[0296] See also Figure 6 , the specific steps for obtaining the implementation effect report are as follows:

[0297] Based on the dose adjustment plan, the dose implementation effect during radiotherapy is monitored in real time using the formula:

[0298] ;

[0299] Calculate and generate actual dose records ;

[0300] in, is the deviation ratio, is the newly added weight parameter;

[0301] Based on the actual dose record, the dynamic data during the treatment process is continuously recorded using the formula:

[0302] ;

[0303] Accumulate the implemented dose data and generate dynamic dose data ;

[0304] in, For the duration of each time point, is the time weighting factor;

[0305] Compare the dynamic dose data with the planned dose using the formula:

[0306] ;

[0307] Calculate the deviation ratio from the planned dose and generate implementation effect report ;

[0308] in, is the planned dose, To report on the implementation effect.

[0309] formula:

[0310] ;

[0311] in:

[0312] : Adjusted theoretical dose;

[0313] : dose deviation ratio;

[0314] : Weight parameters adjusted from clinical data;

[0315] : Adjustment coefficient for the target condition;

[0316] Calculation example:

[0317] set up , , .

[0318] Calculation process:

[0319] ;

[0320] Result analysis:

[0321] It is the actual dose delivered. 250 Gray represents the actual dose delivered after adjustment for reference deviation and target condition.

[0322] formula:

[0323] ;

[0324] in:

[0325] : The duration of each time point;

[0326] : Time weighting factor

[0327] Calculation example:

[0328] There are three time points, each of which casts 250 Gray. The duration of each time point is They are hours, and the weighting factor They are .

[0329] Calculation process:

[0330] ;

[0331] Result analysis:

[0332] It is the actual cumulative dose delivered, and 1100 Gray represents the total dose delivered during the entire treatment process weighted by reference time.

[0333] formula:

[0334] ;

[0335] in:

[0336] : The actual cumulative dose administered;

[0337] : planned dose;

[0338] Calculation example:

[0339] set up From the previous result, we know .

[0340] Calculation process:

[0341] ;

[0342] Result analysis:

[0343] It is the deviation ratio from the planned dose, and 0.316 represents the square root ratio of the dose deviation, which can be used to evaluate the consistency of the treatment effect with the expected one.

[0344] The above are only preferred embodiments of the present invention and are not intended to limit the present invention in other forms. Any technician familiar with the profession may use the technical contents disclosed above to change or modify them into equivalent embodiments with equivalent changes and apply them to other fields. However, any simple modification, equivalent change and modification made to the above embodiments based on the technical essence of the present invention without departing from the technical solution of the present invention still falls within the protection scope of the technical solution of the present invention.

Claims

1. A radiation dose adjustment system based on deep learning, characterized in that: The system comprises: The image analysis module is based on the input of medical image data, scans and distinguishes the boundary between tumors and normal tissues, refines the size and shape of tumors in images through edge detection, extracts key features in images through contrast enhancement, and generates tumor feature data; The steps for obtaining the tumor characteristic data are specifically as follows: For the input medical image data, Gaussian smoothing is performed using the formula: ； Generate smoothed image ; in, represents the image after Gaussian smoothing. represents the Gaussian kernel, which is used for smoothing to reduce image noise and details. Represents the input medical image, which is the raw data processed by Gaussian smoothing; Based on the smoothed image, the Canny algorithm is used for edge detection, using the formula: ； Generate edge enhancement map ; in, represents the edge enhancement map, represents the Sobel operator in the x direction, represents the Sobel operator in the y direction; The edge enhancement image is enhanced using a contrast enhancement algorithm, using the formula: ； Generate contrast enhancement image ; in, represents the contrast enhancement map, Represents the contrast adjustment factor, which is used to adjust the contrast of the image to show edges and details more clearly. and They represent the maximum and minimum values ​​in the edge enhancement map, respectively, and are used to normalize and enhance contrast; Extract and summarize key features from the contrast-enhanced image, combined with the size and morphology information of the tumor, using the formula: ； Generating tumor characterization data ; in, Represents tumor characteristic data, Representative The weight of a feature reflects the importance of the feature in tumor diagnosis. Contrast enhancement image The The weighted value of the feature, Represents the weighted reinforcement index of the feature, which is used to further adjust the influence of the feature; The clinical parameter analysis module is based on the tumor characteristic data, collects the patient's body position information, tumor type and previous radiation reaction records, analyzes and calculates the tumor's radiosensitivity index, and obtains the clinical parameter coordination result; The steps for obtaining the clinical parameter synergy results are specifically as follows: The tumor characteristic data and the patient's body position information are integrated using the formula: ； Computational generation of the initial synthetic dataset ; in, represents the initial comprehensive dataset, and Both represent adjustment coefficients, which are used to adjust tumor characteristic data , cosine transform of body position information and the square of the position information Contribution to the initial comprehensive data set to reflect the weight of different factors on the comprehensive results; The initial comprehensive data were combined with tumor type using the formula: ； Generate intermediate dataset ; in, represents the intermediate data set, and Represents the weight coefficient, used to adjust The square and tumor type The effect of the square root treatment on the intermediate data set. These coefficients reflect the importance of different data transformations on the results. With reference to the intermediate data set and the previous radiation response records, the formula is used: ； Radiosensitivity index ; in, represents the radiosensitivity index, Represents the coefficient for adjusting the sensitivity of radiation response, affecting Attenuation effects and sensitivity thresholds in calculations The intensity of the reaction changes beyond the previous It reflects the exponential decay of the radiation reaction. Standardization for adjusting radiosensitivity indicators; The dose decision module is based on the clinical parameter synergy results and compared with the existing radiation dose database. By analyzing the correlation between pathological characteristics and clinical data, it adjusts the individualized dose distribution, refines the matching degree and adaptability of the dose, and obtains the dose adjustment plan; The steps for obtaining the correlation between the pathological features and the clinical data are specifically as follows: Based on the clinical parameters synergistic results and the existing radiation dose database, the formula was used: ； Calculate the initial relevance score ; in, represents the initial correlation score, which is used to preliminarily evaluate the relationship between pathological features and radiation dose; represents the radiosensitivity index, in which the square is used to amplify the influence to emphasize its importance in the score; represents the weight factor of each sample, which is used to adjust its influence in the calculation according to the characteristics of the sample; represents the number of samples to ensure averaging; The initial relevance score is iteratively analyzed using the formula: ； Perform normalization to fit the different data ranges and generate standardized scores ; in, Represents a standardized score, which is used to Converted into a unified comparison standard, and Respectively represent The minimum and maximum values ​​of , ensuring that it is distributed within a fixed range, represents a small constant, added to avoid the case where the denominator is zero; Using the standardized scores and the variance of the clinical dataset, the formula was used: ； Calculate the Pearson correlation coefficient to determine the correlation between pathological features and clinical data ; in, Represents the Pearson correlation coefficient between pathological features and clinical data, which is used to measure the linear correlation between the two. Represents specific clinical data for use with Conduct correlation analysis, and Respectively represent and The mean of , which is used to center the data during the calculation process; The steps for obtaining the dosage adjustment plan are specifically as follows: Based on the correlation between the pathological features and clinical data, the initial dose adjustment was performed using the formula: ； Calculate preliminary dose adjustments ; Among them, among them, represents the initial dose adjustment value, stands for basal dose, which is the original amount before dose adjustment. represents the adjustment coefficient, which is used to calculate the correlation between pathological features and clinical data Adjust the dose, including Used to emphasize the degree of deviation of the correlation from the baseline (0.5) and to adjust the dose in a nonlinear manner; The preliminary dose adjustment value is refined with reference to individual differences among patients, using the formula: ； Adjust the dose to match individual response and get dose adjustment plan ; in, represents the final dose adjusted to match individual response, Represents the individual difference adjustment coefficient, which is processed according to the square root of the intensity of the patient's side effects , and this coefficient reflects the impact of individual sensitivity and reaction intensity on dosage requirements, making dosage adjustments more accurately adapted to individual differences; The dose implementation monitoring module monitors the dose implementation effect during the radiotherapy process in real time based on the dose adjustment scheme, records the dynamic data of the treatment, adjusts the dose deviation during the treatment, and generates an implementation effect report; The specific steps for obtaining the implementation effect report are as follows: Based on the dose adjustment scheme, the dose implementation effect during radiotherapy is monitored in real time using the formula: ； Calculate and generate actual dose records ; in, Represents the actual dose record, which is used to adjust the dose in the final step. Based on the deviation ratio And the newly added weight parameter and The actual dose is adjusted to Reflects the basic deviation rate, and Indicates the additional weighting factors affecting the dose; Based on the actual dose record, the dynamic data during the treatment process is continuously recorded using the formula: ； Accumulate the implemented dose data and generate dynamic dose data ; in, Represents the cumulative real-time dose data, which is obtained by continuously recording the dynamic data during the treatment process and weighted accumulation. and Respectively represent the duration and time weighting factor of each time point, which are used to calculate the cumulative dose in each treatment period; Compare the dynamic dose data with the planned dose using the formula: ； Calculate the deviation ratio from the planned dose and generate an implementation effect report ; in, Representative implementation effect report, by calculating dynamic dose data With planned dose The deviation ratio between them is obtained.

Images