Bladder EIT three-dimensional shape real-time reconstruction and capacity measurement method based on deep learning

Abstract

The invention provides a bladder EIT three-dimensional shape real-time reconstruction and capacity measurement method based on deep learning, and belongs to the technical field of deep learning, and the method comprises the steps: generating a three-dimensional abdomen simulation model according to a collected three-dimensional CT image, and constructing a bladder EIT data set containing abnormal conditions; acquiring bladder position information according to the three-dimensional abdomen simulation model to construct a loss function containing position regularization, taking EIT measured values in the bladder EIT data set as input of a deep learning network, establishing a mapping relation between the EIT measured values and bladder morphology, and storing a trained network model; and taking the EIT measurement value of the actual patient as the input of the trained network model, performing real-time comparative analysis on the obtained bladder capacity of the patient and the bladder capacity calculated by the CT true value of the patient in the radiotherapy process, and performing real-time optimization and correction on the bladder capacity of the patient. The three-dimensional shape of the bladder can be accurately reconstructed without depending on the traditional image technology, and the measurement of the bladder capacity is optimized in real time.

Description

Technical Field

[0001] The present invention relates to the field of deep learning technology, and in particular to a method for real-time reconstruction of bladder EIT three-dimensional morphology and capacity measurement based on deep learning. Background Art

[0002] Bladder morphology reconstruction and capacity measurement play an important role in precision radiotherapy, especially in lower abdominal radiotherapy. Changes in bladder morphology and capacity directly affect the positioning of the treatment target and dose distribution. Changes in bladder filling status may cause adjacent normal tissues to be exposed to radiation and increase the risk of side effects. Therefore, real-time monitoring of bladder morphology and capacity is crucial to ensure accurate positioning of the target and reduce radiation damage to healthy tissues.

[0003] Traditional methods for measuring bladder morphology and capacity mainly rely on clinical operations and static images, such as ultrasound, CT or MRI. Although these methods can provide relatively accurate capacity information, there are still certain measurement errors due to factors such as patient position, bladder morphology changes and measurement errors. Ultrasound is a commonly used method for bladder capacity assessment, with the advantages of non-invasiveness, real-time and low cost. However, ultrasound imaging has low accuracy in measuring bladder morphology and volume, which is limited by factors such as the operator's technical level, changes in patient position and visibility of the bladder wall. In addition, ultrasound measurement usually needs to be performed under static conditions, which is difficult to cope with the dynamic changes of the bladder during treatment. CT and MRI imaging can provide high-resolution bladder images, which can more accurately reflect the bladder morphology and volume. These methods are usually combined with precise positioning of the tumor target area to provide detailed anatomical data for radiotherapy. However, a significant disadvantage of CT and MRI is that they cannot capture the dynamic changes of the bladder in real time, especially during radiotherapy, when the shape and capacity of the bladder will change with changes in physiological processes such as urination and water intake.

[0004] Electrical impedance tomography (EIT) is an imaging technique based on the principle of bioimpedance, which reconstructs images by measuring the difference in electrical impedance of different tissues. The main advantages of EIT are its non-invasiveness, real-time nature, and ability to measure in dynamic situations. Unlike CT and MRI, EIT can provide continuous and updated information on bladder morphology and capacity, and is particularly promising for dynamic bladder assessment. Despite its great potential, EIT still faces some technical challenges in practical applications. First, due to the small difference in electrical impedance between fluid and tissue in the bladder, the resolution of electrical impedance imaging is low, which makes it difficult to accurately reconstruct the morphology and capacity of the bladder in complex clinical scenarios. Second, EIT imaging technology requires data acquisition on the abdominal surface through multiple electrode arrays, but due to the thickness of the abdominal wall, tissue type, and patient size, the EIT signal may be interfered, affecting the imaging accuracy.

[0005] Therefore, it is necessary to design a deep learning-based method for bladder EIT three-dimensional morphology reconstruction and capacity measurement. Summary of the invention

[0006] The present invention provides a method for real-time reconstruction of bladder EIT three-dimensional morphology and capacity measurement based on deep learning, which is used to combine EIT technology and deep learning algorithm, and can accurately reconstruct the three-dimensional morphology of the bladder without relying on traditional imaging technology, and optimize the measurement of bladder capacity in real time.

[0007] The present invention provides a method for real-time reconstruction of bladder EIT three-dimensional morphology and capacity measurement based on deep learning, comprising: S1: Based on the collected 3D CT images, the contour point information of the lower abdomen, bladder, pelvis and subcutaneous fat is extracted to generate a 3D abdominal simulation model. At the same time, based on the pre-deployed flexible electrodes, the EIT measurement values ​​under normal conditions, electrode drop and noise are collected to construct a bladder EIT dataset containing abnormal conditions. S2: Obtain bladder position information based on the three-dimensional abdominal simulation model to construct a loss function containing position regularization, and use the EIT measurement values ​​in the bladder EIT dataset as the input of the deep learning network to establish a mapping relationship between the EIT measurement values ​​and the bladder morphology, output the three-dimensional morphological image of the bladder, and save the trained network model; S3: Calculate the number of voxels of the bladder in the reconstructed three-dimensional topographic image, and combine the relationship between each voxel and the real physical size to obtain the reconstructed bladder capacity; S4: The actual patient's EIT measurement value is used as the input of the trained network model, and the obtained patient's bladder capacity is compared and analyzed in real time with the bladder capacity calculated by the patient's CT true value during radiotherapy, so as to optimize and correct the patient's bladder capacity in real time.

[0008] Preferably, the collected three-dimensional CT images and bladder EIT data sets are divided into a training set, a validation set and a test set in a ratio of 8:1:1, respectively.

[0009] Preferably, before saving the trained network model, the method further includes: Establish a deep learning network structure based on V-Net; On the basis of the network structure, a fully connected layer and SE attention mechanism are added, and the cross entropy loss function integrating position prior information and the loss function including position regularization are combined to obtain a network model; The network model is trained using the training set, and at the same time, the verification error of the verification set on the trained network model is monitored. When the verification error is in an increasing state, the training of the network model is stopped and saved.

[0010] Preferably, obtaining the reconstructed bladder capacity includes:

[0011] in, represents the bladder capacity after reconstruction; is the volume of each voxel, The voxel value of the volume unit (i, j, k) representing the voxel indicates that the voxel is bladder tissue; Indicates the horizontal, vertical and vertical quantities of the volume unit.

[0012] Preferably, the patient's bladder capacity is optimized and corrected in real time, including: Obtain the absolute value difference between the patient's bladder capacity and the bladder capacity calculated from the patient's CT true value during radiotherapy; If the absolute value difference is greater than a preset threshold, the patient's bladder capacity is weightedly fused with the bladder capacity calculated based on the CT true value to obtain the required capacity, thereby achieving real-time optimization correction;

[0013] in, is the required capacity; represents the weighting factor; Indicates the patient's bladder capacity; Indicates the bladder capacity calculated based on the true CT value.

[0014] Preferably, the method further comprises: determining a weighting factor, specifically comprising: Extracting absolute value differences of N1 actual patients, and counting a first number M1 of the N1 absolute value differences that are greater than a preset threshold and a second number M2 that are less than or equal to the preset threshold; The absolute value differences of the first quantity M1 are sorted and analyzed discretely according to the historical measurement time of the actual patient, and the absolute value differences of the second quantity M2 are sorted and analyzed continuously according to the historical measurement time of the actual patient; Determine the degree of dispersion of each absolute value difference in the first quantity M1 according to the discrete analysis result, calculate the average degree of the remaining M1-1 absolute value differences after dividing the first quantity M1, and obtain the residual of the corresponding absolute value difference; Sort all absolute value differences in the first quantity M1 by size according to the residuals to obtain a first sorting sequence number; Perform consistency analysis on the first sorting sequence number and the second sorting sequence number whose absolute value differences in the first quantity M1 are sorted according to the historical measurement time to determine the first sequence number change quantity NB, wherein the consistency analysis means that the sequence number of each absolute value difference in the first sorting sequence number and the sequence number of the corresponding absolute value difference in the second sorting sequence number are considered consistent within the range of (-a1, a1), wherein a1 represents the change quantity; Determine a stability coefficient of the absolute value difference of the second number M2 according to the continuous analysis result, and calculate the product of the stability coefficient and the average value of the absolute value difference of the second number M2; Determine a first difference absolute value between each absolute value difference in the second quantity M2 and the product, and a second difference absolute value between the product and a preset threshold, and obtain a current average value of the first difference absolute value and the second difference absolute value; Calculate the variance of all current means , if the variance Not less than the preset value , then keep the second number M2 unchanged, and obtain the weighting factor based on the first sequence number change number NB, the second number M2, and the first number M1 ; If the variance Less than the preset value , the second quantity M2 is adjusted to obtain the third quantity M3, and based on the first sequence number change quantity NB, the third quantity M3, and the first quantity M1, a weighting factor is obtained. ; .

[0015] Preferably, adjusting the second quantity M2 to obtain the third quantity M3 includes:

[0016] in, Indicates the round-up symbol.

[0017] Preferably, generating a three-dimensional abdomen simulation model includes: Annotating contour points of different parts of each three-dimensional CT image to obtain an annotation map, and performing center calibration according to the closed contour line of each part in the annotation map to obtain the center point of each part; The center points in the same three-dimensional CT image are sequentially connected according to the structural order of the lower abdomen, bladder, pelvis and subcutaneous fat to obtain a first connection line; Each first connection line is standardized according to the size mapping relationship with the body frame of the collected object to obtain a first standard line, and all first standard lines are aligned and placed with the center point of a random part as the alignment reference point, and the center deviation set of each remaining part is obtained to obtain the deviation coefficient;

[0018] in, represents the deviation coefficient of the x0th remaining part; represents the point density of the e1th unit area obtained after dividing the center deviation set of the x0th residual part into unit areas; Indicates all The maximum value in ; Representation and The area of ​​the corresponding unit region; represents the area occupied by the center deviation set of the x0th residual part; represents the center coordinate point of the e1th unit area based on the center deviation set of the x0th residual part; Representation and The center coordinate point of the corresponding unit area; Respectively based on all ,all The variance of All deviation coefficients are averaged to obtain an average coefficient, and an image enhancement coefficient is determined from a coefficient-enhancement comparison table, and image enhancement processing is performed on a three-dimensional CT image whose image clarity is less than a preset clarity according to the image enhancement coefficient to obtain an enhanced image; The contour point information of the lower abdomen, bladder, pelvis and subcutaneous fat is extracted from each enhanced image and the three-dimensional CT image with an image clarity not less than a preset clarity to generate a three-dimensional abdominal simulation model.

[0019] Compared with the prior art, the present invention has the following beneficial effects: Combining EIT technology and deep learning algorithms, it is possible to accurately reconstruct the three-dimensional morphology of the bladder and optimize the measurement of bladder capacity in real time without relying on traditional imaging technology.

[0020] Other features and advantages of the present invention will be described in the following description, and partly become apparent from the description, or understood by practicing the present invention. The purpose and other advantages of the present invention can be realized and obtained by the structures particularly pointed out in the written description and the accompanying drawings.

[0021] The technical solution of the present invention is further described in detail below through the accompanying drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings: Figure 1 This is a flow chart of a method for real-time reconstruction of bladder EIT three-dimensional morphology and capacity measurement based on deep learning in an embodiment of the present invention; Figure 2 A diagram of a three-dimensional abdomen simulation model in an embodiment of the present invention; Figure 3 Schematic diagram of the structure of the network model in the embodiment of the present invention. DETAILED DESCRIPTION

[0023] The preferred embodiments of the present invention are described below in conjunction with the accompanying drawings. It should be understood that the preferred embodiments described herein are only used to illustrate and explain the present invention, and are not used to limit the present invention.

[0024] The present invention provides a method for real-time reconstruction of bladder EIT three-dimensional morphology and capacity measurement based on deep learning, such as Figure 1 As shown, including: S1: Based on the collected 3D CT images, the contour point information of the lower abdomen, bladder, pelvis and subcutaneous fat is extracted to generate a 3D abdominal simulation model. At the same time, based on the pre-deployed flexible electrodes, the EIT measurement values ​​under normal conditions, electrode drop and noise are collected to construct a bladder EIT dataset containing abnormal conditions. S2: Obtain bladder position information based on the three-dimensional abdominal simulation model to construct a loss function containing position regularization, and use the EIT measurement values ​​in the bladder EIT dataset as the input of the deep learning network to establish a mapping relationship between the EIT measurement values ​​and the bladder morphology, output the three-dimensional morphological image of the bladder, and save the trained network model; S3: Calculate the number of voxels of the bladder in the reconstructed three-dimensional topographic image, and combine the relationship between each voxel and the real physical size to obtain the reconstructed bladder capacity; S4: The actual patient's EIT measurement value is used as the input of the trained network model, and the obtained patient's bladder capacity is compared and analyzed in real time with the bladder capacity calculated by the patient's CT true value during radiotherapy, so as to optimize and correct the patient's bladder capacity in real time.

[0025] Preferably, the collected three-dimensional CT images and bladder EIT data sets are divided into a training set, a validation set and a test set in a ratio of 8:1:1, respectively.

[0026] Preferably, before saving the trained network model, the method further includes: Establish a deep learning network structure based on V-Net; On the basis of the network structure, a fully connected layer and SE attention mechanism are added, and the cross entropy loss function integrating position prior information and the loss function including position regularization are combined to obtain a network model; The network model is trained using the training set, and at the same time, the verification error of the verification set on the trained network model is monitored. When the verification error is in an increasing state, the training of the network model is stopped and saved.

[0027] Preferably, obtaining the reconstructed bladder capacity includes:

[0028] in, represents the bladder capacity after reconstruction; is the volume of each voxel, The voxel value of the volume unit (i, j, k) representing the voxel indicates that the voxel is bladder tissue; Indicates the horizontal, vertical and vertical quantities of the volume unit.

[0029] Preferably, the patient's bladder capacity is optimized and corrected in real time, including: Obtain the absolute value difference between the patient's bladder capacity and the bladder capacity calculated from the patient's CT true value during radiotherapy; If the absolute value difference is greater than a preset threshold, the patient's bladder capacity is weightedly fused with the bladder capacity calculated based on the CT true value to obtain the required capacity, thereby achieving real-time optimization correction;

[0030] in, is the required capacity; represents the weighting factor; Indicates the patient's bladder capacity; Indicates the bladder capacity calculated based on the true CT value.

[0031] In this embodiment, before generating the three-dimensional abdomen simulation model, the conductivity attribute of each data point scanned in the three-dimensional CT image is annotated to more realistically simulate the electrical impedance characteristics of various tissues in the human body. According to the actual conductivity distribution, the conductivity of the bladder is set to 0.5 S / m, the conductivity of the pelvis is 0.05 S / m, the conductivity of the subcutaneous fat is 0.2S / m, and the conductivity of other parts is uniformly set to 0.1 S / m. For the three-dimensional abdomen simulation model, Figure 2 shown.

[0032] In this embodiment, before constructing the bladder EIT data set including abnormal conditions, the following steps are performed: 16 flexible electrodes are arranged in a half circle about 8 cm below the navel to collect EIT measurement data under different conditions, including normal conditions, electrode drop conditions, and noise conditions. This enables the training data set to have stronger adaptability and cover various possible abnormal conditions.

[0033] In this embodiment, a deep learning network structure based on V-Net is used. V-Net is a convolutional neural network commonly used in medical image segmentation. Its structure has strong spatial information processing capabilities and is suitable for bladder morphology reconstruction. Figure 3 As shown in the figure, in order to improve the network's ability to capture details, a fully connected layer (FC) is added before the V-Net network to extract features from the original image and strengthen the global information of feature learning as follows:

[0034] Among them, X is the input feature map, z is the global average pooling feature of each channel, W1 and W2 are the weights of the two-layer fully connected network, is the Sigmoid activation function, Represents an element-by-element multiplication operation. By learning the weights for each channel, different importance is assigned to each channel, thereby improving the representation ability of the model.

[0035] Loss function design: The loss function uses the cross entropy loss function (CrossEntropy Loss) that integrates the position prior information, which is often used to measure the distance between the target and the prediction result in classification tasks. The public statement is as follows:

[0036] Cross entropy loss can help the network optimize classification accuracy and has a good effect on bladder morphology reconstruction tasks. Through the optimization of the cross entropy loss function, the network can more accurately learn the segmentation boundaries of each bladder area and reconstruct a more accurate three-dimensional bladder morphology.

[0037] At the same time, the bladder position prior information is introduced into the loss function, and the bladder center position predicted by the model is optimized by imposing constraints. According to the approximate center coordinates (x0, y0) of the bladder provided by the CT image, a regularization term can be added to make the bladder center output by the model close to the prior center:

[0038] Among them, x pred and pred is the coordinate of the bladder center predicted by the model, and λ is the regularization coefficient used to control the impact of this term on the overall loss.

[0039] Finally, the loss function of the network can combine the standard cross entropy loss and the loss of position prior information to form a comprehensive loss function:

[0040] During the training process, the model is trained using the training set, and the performance of the model is monitored in real time using the validation set. If the error on the validation set starts to rise, it means that the model may be overfitting. In this case, the training should be stopped immediately and the current optimal network model should be saved. The optimization algorithm used for network training is the Adam optimizer, and a reasonable learning rate scheduling strategy is set.

[0041] The beneficial effect of the above technical solution is that by combining EIT technology and deep learning algorithms, it is possible to accurately reconstruct the three-dimensional morphology of the bladder and optimize the measurement of bladder capacity in real time without relying on traditional imaging technology.

[0042] The present invention provides a method for real-time reconstruction of bladder EIT three-dimensional morphology and capacity measurement based on deep learning, which also includes: determining a weighting factor, specifically including: Extracting absolute value differences of N1 actual patients, and counting a first number M1 of the N1 absolute value differences that are greater than a preset threshold and a second number M2 that are less than or equal to the preset threshold; The absolute value differences of the first quantity M1 are sorted and analyzed discretely according to the historical measurement time of the actual patient, and the absolute value differences of the second quantity M2 are sorted and analyzed continuously according to the historical measurement time of the actual patient; Determine the degree of dispersion of each absolute value difference in the first quantity M1 according to the discrete analysis result, calculate the average degree of the remaining M1-1 absolute value differences after dividing the first quantity M1, and obtain the residual of the corresponding absolute value difference; Sort all absolute value differences in the first quantity M1 by size according to the residuals to obtain a first sorting sequence number; Perform consistency analysis on the first sorting sequence number and the second sorting sequence number whose absolute value differences in the first quantity M1 are sorted according to the historical measurement time to determine the first sequence number change quantity NB, wherein the consistency analysis means that the sequence number of each absolute value difference in the first sorting sequence number and the sequence number of the corresponding absolute value difference in the second sorting sequence number are considered consistent within the range of (-a1, a1), wherein a1 represents the change quantity; Determine a stability coefficient of the absolute value difference of the second number M2 according to the continuous analysis result, and calculate the product of the stability coefficient and the average value of the absolute value difference of the second number M2; Determine a first difference absolute value between each absolute value difference in the second quantity M2 and the product, and a second difference absolute value between the product and a preset threshold, and obtain a current average value of the first difference absolute value and the second difference absolute value; Calculate the variance of all current means , if the variance Not less than the preset value , then keep the second number M2 unchanged, and obtain the weighting factor based on the first sequence number change number NB, the second number M2, and the first number M1 ; If the variance Less than the preset value , the second quantity M2 is adjusted to obtain the third quantity M3, and based on the first sequence number change quantity NB, the third quantity M3, and the first quantity M1, a weighting factor is obtained. ; .

[0043] Preferably, adjusting the second quantity M2 to obtain the third quantity M3 includes:

[0044] in, Indicates the round-up symbol.

[0045] In this embodiment, the actual absolute value difference of the patient refers to the absolute value difference between the patient's bladder capacity and the bladder capacity calculated by the patient's CT true value during radiotherapy, and N1 is greater than 200.

[0046] In this embodiment, the preset threshold is pre-set and the value is 30 ml.

[0047] In this embodiment, the historical measurement time refers to the time when the model is used for measurement. There is a sequence of measurements, so sorting can be achieved based on the measurement time.

[0048] In this embodiment, the discrete analysis result refers to the discrete degree of each absolute value difference in the first quantity M1 obtained by averaging the absolute value of the difference between the absolute value difference and each remaining absolute value difference in the first quantity M1.

[0049] In this embodiment, the residual error=the dispersion degree of the corresponding absolute value difference in the first number M1 minus the residual M1 divided by itself minus the average degree of 1 absolute value difference.

[0050] In this embodiment, for example, the second sorting sequence numbers sorted according to the historical measurement time are: u1--1, u2--2, u4--3, u7--4, and the first sorting sequence numbers are: u2--1, u4--2, u1--3, u7--4.

[0051] Among them, the serial number of u7 remains unchanged, the serial number of u4 moves 1 bit to the left, the serial number of u2 moves 1 bit to the left, and the serial number of u1 moves 2 bits to the right. At this time, assuming that the value of a1 is 1, that is, only u1 meets the requirements. At this time, the first serial number change quantity NB is 1.

[0052] In this embodiment, -a1 represents a shift to the left by a1 bit, a1 represents a shift to the right by a1 bit, and a1 is a positive integer.

[0053] In this embodiment, the stability coefficient=1-(maximum value of the absolute value difference of the second quantity M2-minimum value of the absolute value difference of the second quantity M2) / fitting coefficient of the absolute value difference of the second quantity M2 determined by continuous analysis results.

[0054] In this embodiment, the product=stability coefficient×average value of the absolute value difference of the second number M2.

[0055] In this embodiment, the current average value=(the absolute value of the first difference+the absolute value of the second difference) / 2.

[0056] The beneficial effect of the above technical solution is: by obtaining N1 absolute value differences as the basis for data analysis, and then subsequently determining the weighting factors in different situations through continuous analysis and discrete analysis, the discrete analysis results are used as the main analysis standard to determine the changes in the sequence numbers, and the product of the stability coefficients under the continuous analysis results is combined to determine the variance for the current average value, and then through intelligent classification judgment, the weighting factors can be reliably obtained, providing a basis for subsequent capacity optimization.

[0057] The present invention provides a method for real-time reconstruction of bladder EIT three-dimensional morphology and capacity measurement based on deep learning, generating a three-dimensional abdominal simulation model, including: Annotating contour points of different parts of each three-dimensional CT image to obtain an annotation map, and performing center calibration according to the closed contour line of each part in the annotation map to obtain the center point of each part; The center points in the same three-dimensional CT image are sequentially connected according to the structural order of the lower abdomen, bladder, pelvis and subcutaneous fat to obtain a first connection line; Each first connection line is standardized according to the size mapping relationship with the body frame of the collected object to obtain a first standard line, and all first standard lines are aligned and placed with the center point of a random part as the alignment reference point, and the center deviation set of each remaining part is obtained to obtain the deviation coefficient;

[0058] in, represents the deviation coefficient of the x0th remaining part; represents the point density of the e1th unit area obtained after dividing the center deviation set of the x0th residual part into unit areas; Indicates all The maximum value in ; Representation and The area of ​​the corresponding unit region; represents the area occupied by the center deviation set of the x0th residual part; represents the center coordinate point of the e1th unit area based on the center deviation set of the x0th residual part; Representation and The center coordinate point of the corresponding unit area; Respectively based on all ,all The variance of All deviation coefficients are averaged to obtain an average coefficient, and an image enhancement coefficient is determined from a coefficient-enhancement comparison table, and image enhancement processing is performed on a three-dimensional CT image whose image clarity is less than a preset clarity according to the image enhancement coefficient to obtain an enhanced image; The contour point information of the lower abdomen, bladder, pelvis and subcutaneous fat is extracted from each enhanced image and the three-dimensional CT image with an image clarity not less than a preset clarity to generate a three-dimensional abdominal simulation model.

[0059] In this embodiment, the first connection lines are connected in order from top to bottom and from left to right.

[0060] In this embodiment, the annotation of contour points in the three-dimensional CT image can be obtained by directly annotating the contour extraction correlation network, which belongs to the prior art. However, in the annotation process, due to the unclear image and other conditions, certain errors may occur in the annotation. Therefore, the center point of each part in the standard image is obtained to draw a connecting line, and then the center deviation set is obtained by alignment.

[0061] In this embodiment, center calibration refers to calibration performed based on the center point of the closed contour line of the corresponding part.

[0062] In this embodiment, since the collected three-dimensional CT images are for different patients and there are certain differences in the ages of the patients, and the sizes of the parts corresponding to patients of different ages are different, it is necessary to rely on the size mapping relationship to standardize the size of the parts to facilitate the subsequent alignment and placement of alignment reference points based on the parts.

[0063] In this embodiment, for example, the center point of the lower abdomen is used as the alignment reference point for alignment and placement to obtain a center deviation set based on the bladder, pelvis, and subcutaneous fat structure, and the center deviation set is a position deviation set for the center point of the corresponding part, so the deviation coefficient can be obtained by analyzing the set.

[0064] In this embodiment, the average coefficient = the average value of all deviation coefficients.

[0065] In this embodiment, the coefficient-enhancement comparison table includes an image enhancement coefficient that matches the average coefficient, and image enhancement processing according to the enhancement coefficient is based on matching in the enhancement coefficient-enhancement method table, which includes image enhancement methods for different enhancement coefficients, and enhancement processing can be directly performed to obtain an enhanced image.

[0066] In this embodiment, the contour point information finally extracted is fused at the same position to obtain a three-dimensional abdomen simulation model.

[0067] The beneficial effects of the above technical solution are: different three-dimensional CT images are annotated with contour points to obtain the center point of each part based on the closed contour line, and then connected in sequence to obtain the first connection line, providing a basis for subsequent alignment processing, and ensuring alignment standardization through adjustment of the size mapping relationship, ensuring the reliability of subsequent deviation coefficient acquisition, providing a real basis for image enhancement, and then realizing the rationality of image enhancement, indirectly improving the accuracy of optimizing bladder capacity measurement.

[0068] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.

Claims

1. A method for real-time reconstruction of bladder EIT three-dimensional morphology and capacity measurement based on deep learning, characterized in that: include: S1: Based on the collected 3D CT images, the contour point information of the lower abdomen, bladder, pelvis and subcutaneous fat is extracted to generate a 3D abdominal simulation model. At the same time, based on the pre-deployed flexible electrodes, the EIT measurement values ​​under normal conditions, electrode drop and noise are collected to construct a bladder EIT dataset containing abnormal conditions. S2: Obtain bladder position information based on the three-dimensional abdominal simulation model to construct a loss function containing position regularization, and use the EIT measurement values ​​in the bladder EIT dataset as the input of the deep learning network to establish a mapping relationship between the EIT measurement values ​​and the bladder morphology, output the three-dimensional morphological image of the bladder, and save the trained network model; S3: Calculate the number of voxels of the bladder in the reconstructed three-dimensional topographic image, and combine the relationship between each voxel and the real physical size to obtain the reconstructed bladder capacity; S4: The actual patient's EIT measurement value is used as the input of the trained network model, and the obtained patient's bladder capacity is compared and analyzed in real time with the bladder capacity calculated by the patient's CT true value during radiotherapy, so as to optimize and correct the patient's bladder capacity in real time.

2. The method for real-time reconstruction of bladder EIT three-dimensional morphology and capacity measurement based on deep learning according to claim 1, characterized in that: The collected three-dimensional CT images and bladder EIT datasets were divided into training set, validation set and test set in the ratio of 8:1:

1.

3. The method for real-time reconstruction of bladder EIT three-dimensional morphology and capacity measurement based on deep learning according to claim 2, characterized in that: Before saving the trained network model, it also includes: Establish a deep learning network structure based on V-Net; On the basis of the network structure, a fully connected layer and SE attention mechanism are added, and the cross entropy loss function integrating position prior information and the loss function including position regularization are combined to obtain a network model; The network model is trained using the training set, and at the same time, the verification error of the verification set on the trained network model is monitored. When the verification error is in an increasing state, the training of the network model is stopped and saved.

4. The method for real-time reconstruction of bladder EIT three-dimensional morphology and capacity measurement based on deep learning according to claim 1, characterized in that: Obtain reconstructed bladder capacity, including: ； in, represents the reconstructed bladder capacity; is the volume of each voxel, The voxel value of the volume unit (i, j, k) representing the voxel indicates that the voxel is bladder tissue; Indicates the horizontal, vertical and vertical quantities of the volume unit.

5. The method for real-time reconstruction of bladder EIT three-dimensional morphology and capacity measurement based on deep learning according to claim 1, characterized in that: Real-time optimization and correction of the patient's bladder capacity, including: Obtain the absolute value difference between the patient's bladder capacity and the bladder capacity calculated from the patient's CT true value during radiotherapy; If the absolute value difference is greater than a preset threshold, the patient's bladder capacity is weightedly fused with the bladder capacity calculated based on the CT true value to obtain the required capacity, thereby achieving real-time optimization correction; ； in, is the required capacity; represents the weighting factor; Indicates the patient's bladder capacity; Indicates the bladder capacity calculated based on the true CT value.

6. The method for real-time reconstruction of bladder EIT three-dimensional morphology and capacity measurement based on deep learning according to claim 5, characterized in that: Also includes: Determine the weighting factors, including: Extracting absolute value differences of N1 actual patients, and counting a first number M1 of the N1 absolute value differences that are greater than a preset threshold and a second number M2 that are less than or equal to the preset threshold; The absolute value differences of the first quantity M1 are sorted and analyzed discretely according to the historical measurement time of the actual patient, and the absolute value differences of the second quantity M2 are sorted and analyzed continuously according to the historical measurement time of the actual patient; Determine the degree of dispersion of each absolute value difference in the first quantity M1 according to the discrete analysis result, calculate the average degree of the remaining M1-1 absolute value differences after dividing the first quantity M1, and obtain the residual of the corresponding absolute value difference; Sort all absolute value differences in the first quantity M1 by size according to the residual to obtain a first sorting sequence number; Perform consistency analysis on the first sorting sequence number and the second sorting sequence number whose absolute value differences in the first quantity M1 are sorted according to the historical measurement time to determine the first sequence number change quantity NB, wherein the consistency analysis means that the sequence number of each absolute value difference in the first sorting sequence number and the sequence number of the corresponding absolute value difference in the second sorting sequence number are considered consistent within the range of (-a1, a1), wherein a1 represents the change quantity; Determine a stability coefficient of the absolute value difference of the second number M2 according to the continuous analysis result, and calculate the product of the stability coefficient and the average value of the absolute value difference of the second number M2; Determine a first difference absolute value between each absolute value difference in the second quantity M2 and the product, and a second difference absolute value between the product and a preset threshold, and obtain a current average value of the first difference absolute value and the second difference absolute value; Calculate the variance of all current means , if the variance Not less than the preset value , then keep the second number M2 unchanged, and obtain the weighting factor based on the first sequence number change number NB, the second number M2, and the first number M1 ; If the variance Less than the preset value , the second quantity M2 is adjusted to obtain the third quantity M3, and based on the first sequence number change quantity NB, the third quantity M3, and the first quantity M1, a weighting factor is obtained. ; 。 7. The method for real-time reconstruction of bladder EIT three-dimensional morphology and capacity measurement based on deep learning according to claim 6, characterized in that: The second quantity M2 is adjusted to obtain a third quantity M3, including: ； in, Indicates the round-up symbol.

8. The method for real-time reconstruction of bladder EIT three-dimensional morphology and capacity measurement based on deep learning according to claim 1, characterized in that: Generate a 3D abdomen simulation model, including: Annotating contour points of different parts of each three-dimensional CT image to obtain an annotation map, and performing center calibration according to the closed contour line of each part in the annotation map to obtain the center point of each part; The center points in the same three-dimensional CT image are sequentially connected according to the structural order of the lower abdomen, bladder, pelvis and subcutaneous fat to obtain a first connection line; Each first connection line is standardized according to the size mapping relationship with the body frame of the collected object to obtain a first standard line, and all first standard lines are aligned and placed with the center point of a random part as the alignment reference point, and the center deviation set of each remaining part is obtained to obtain the deviation coefficient; ； in, represents the deviation coefficient of the x0th remaining part; represents the point density of the e1th unit area obtained after dividing the center deviation set of the x0th residual part into unit areas; Indicates all The maximum value in ; Representation and The area of ​​the corresponding unit region; represents the area occupied by the center deviation set of the x0th residual part; represents the center coordinate point of the e1th unit area based on the center deviation set of the x0th residual part; Representation and The center coordinate point of the corresponding unit area; Respectively based on all ,all The variance of All deviation coefficients are averaged to obtain an average coefficient, and an image enhancement coefficient is determined from a coefficient-enhancement comparison table, and image enhancement processing is performed on a three-dimensional CT image whose image clarity is less than a preset clarity according to the image enhancement coefficient to obtain an enhanced image; The contour point information of the lower abdomen, bladder, pelvis and subcutaneous fat is extracted from each enhanced image and the three-dimensional CT image whose image clarity is not less than the preset clarity to generate a three-dimensional abdominal simulation model.

Images