Intelligent prostate volume measurement method and system based on multi-modal images

Abstract

The application discloses a prostate volume intelligent measurement method and system based on multi-modal images, and belongs to the technical field of medical images. The method comprises the following steps: acquiring an ultrasound image and a nuclear magnetic resonance image of a prostate region of a patient; performing image preprocessing and registration based on the ultrasound image and the nuclear magnetic resonance image; fusing the ultrasound image and the nuclear magnetic resonance image according to a registration result to generate an enhanced image; performing prostate region segmentation by using a deep learning model based on the enhanced image; calculating a prostate volume according to a segmentation result; and generating a diagnosis report containing the prostate volume measurement result. By means of multi-modal image fusion, the method makes full use of the real-time performance of ultrasound imaging and the high-resolution characteristics of nuclear magnetic resonance imaging, effectively overcomes the limitations of single-modal imaging, improves the accuracy of prostate boundary recognition, and enhances the adaptability of the method to different imaging conditions.

Description

TECHNICAL FIELD

[0001] The present application relates to a multi-modal image-based intelligent prostate volume measurement method and system, belonging to the field of medical imaging technology. BACKGROUND

[0002] Prostate diseases are common health problems among middle-aged and elderly men, and accurate measurement of prostate volume is crucial for diagnosis and treatment. Traditional methods of measuring prostate volume mainly rely on single modal medical images, such as ultrasound or magnetic resonance imaging. However, these methods each have limitations, making it difficult to balance measurement accuracy and efficiency.

[0003] Ultrasound imaging is widely used in prostate volume measurement due to its real-time, non-invasive and relatively low-cost characteristics. However, the quality of ultrasound images is often affected by factors such as operator experience, device performance and patient size, resulting in less than ideal repeatability and accuracy of measurement results. In particular, in cases where the prostate boundary is unclear or there is calcification, the error of ultrasound measurement can significantly increase.

[0004] On the other hand, magnetic resonance imaging can provide high-resolution soft tissue contrast, which is beneficial for accurate identification of prostate boundaries. However, magnetic resonance examinations are time-consuming and costly, and are not suitable for some patients (such as those with metal implants in their bodies). In addition, the analysis and volume calculation of magnetic resonance images usually require manual operation by professionals, which not only consumes time and effort, but also may introduce subjective errors.

[0005] In recent years, with the development of computer vision and artificial intelligence technology, some automated prostate volume measurement methods have been proposed. These methods are usually based on deep learning algorithms, such as convolutional neural networks, to achieve automatic segmentation of the prostate region and volume calculation. Although these methods have improved the efficiency of measurement to some extent, there are still some problems. First, most methods still rely on single modal images, and cannot fully utilize the complementary advantages of different imaging techniques. Second, existing deep learning models are often sensitive to changes in image quality and imaging conditions, and their robustness in clinical applications needs to be improved. Finally, these methods often lack effective measurement result verification and optimization mechanisms, making it difficult to ensure reliable measurement results in different situations. SUMMARY

[0006] To improve the accuracy and efficiency of prostate volume measurement, the present application provides a multi-modal image-based intelligent prostate volume measurement method and system, the technical solution being as follows:

[0007] The multi-modal image-based intelligent prostate volume measurement method of the present application comprises:

[0008] Step 1: Obtain the ultrasound image and the magnetic resonance image of the prostate region of the subject and pre-process them;

[0009] Step 2: Register the pre-processed ultrasound image and the magnetic resonance image, and the registration process includes:

[0010] Step 21: Construct an adaptive similarity measurement function to calculate the similarity of the ultrasound image and the magnetic resonance image, and the calculation method is:

[0011] S = a · MI(I1, I2) + (1-a) · SSIM(I1, I2)

[0012] Where MI represents mutual information, SSIM represents structural similarity index, a is a weight coefficient, and I1 and I2 represent two images to be registered respectively;

[0013] Step 22: Take the adaptive similarity measurement function as the objective function, and use the particle swarm optimization algorithm to search for the optimal registration parameters;

[0014] Step 23: Based on the optimal registration parameters, perform spatial alignment on the ultrasound image and the magnetic resonance image;

[0015] Step 3: According to the registration result, fuse the ultrasound image and the magnetic resonance image to generate an enhanced image;

[0016] Step 4: Based on the enhanced image, use a deep learning model to segment the prostate region;

[0017] Step 5: According to the segmentation result, calculate the prostate volume, perform multiple measurements to obtain a volume measurement sequence, and perform statistical analysis on the volume measurement sequence to obtain a prostate volume estimate.

[0018] Optionally, the step 5 further includes volume measurement optimization, and the optimization process includes:

[0019] Step 51: Set a volume measurement error threshold and a maximum number of measurements;

[0020] Step 52: Calculate the relative error of the adjacent two measurement results in the volume measurement sequence, and the calculation formula is:

[0021]

[0022] Where V i is the current measurement value, and V i-1 is the last measurement value;

[0023] Step 53: When the relative error is less than the volume measurement error threshold or the maximum number of measurements is reached, stop measuring;

[0024] Step 54: remove outliers;

[0025] Step 55: calculate the weighted average of the remaining measurement values as the final volume estimate.

[0026] Optionally, the step 4 uses an improved U-Net model as the deep learning segmentation model, uses ResNet18 as the encoder of the U-Net model, and introduces a batch normalization layer in each residual block of ResNet18; and introduces a spatial attention mechanism in each encoding block and decoding block of the U-Net model.

[0027] Optionally, the method uses a cross-modal loss function to train the deep learning model, and the cross-modal loss function is:

[0028] L=αL dice +βL focal +γL boundary

[0029] wherein, L dice is the Dice loss, L focal is the Focal loss, L boundary is the boundary loss, which is used to improve the accuracy of the segmentation boundary, and alpha, beta and gamma are weight coefficients.

[0030] The prostate volume intelligent measurement system based on multi-modal images of the application has the characteristics that the system comprises:

[0031] An image acquisition module is configured to acquire ultrasound images and magnetic resonance images of a prostate region of a patient.

[0032] An image preprocessing module is configured to preprocess the ultrasound images and the magnetic resonance images.

[0033] An image registration module is configured to perform spatial alignment of the ultrasound images and the magnetic resonance images.

[0034] The registration process comprises:

[0035] An adaptive similarity measurement function is constructed to calculate the similarity of the ultrasound images and the magnetic resonance images, and the calculation method is as follows:

[0036] S=α·MI(I1,I2)+(1-α)·SSIM(I1,I2)

[0037] wherein, MI represents mutual information, SSIM represents structural similarity index, alpha is a weight coefficient, and I1 and I2 represent two images to be registered.

[0038] The adaptive similarity measurement function is used as an objective function, and a particle swarm optimization algorithm is used to search for optimal registration parameters.

[0039] align the ultrasound image and the magnetic resonance image in space based on the optimal registration parameter;

[0040] an image fusion module configured to fuse the registered ultrasound image and the magnetic resonance image to generate an enhanced image;

[0041] an image segmentation module configured to perform prostate region segmentation on the enhanced image;

[0042] a volume calculation module configured to calculate a prostate volume according to the segmentation result, perform multiple measurements to obtain a volume measurement sequence, and statistically analyze the volume measurement sequence to obtain a prostate volume estimate.

[0043] Optionally, the system further comprises a measurement optimization module, and the optimization process comprises:

[0044] setting a volume measurement error threshold and a maximum number of measurements;

[0045] calculating a relative error of adjacent measurement results in the volume measurement sequence, and the calculation formula is:

[0046]

[0047] wherein, V i is a current measurement value, V i-1 is a last measurement value;

[0048] stopping the measurement when the relative error is less than the volume measurement error threshold or the maximum number of measurements is reached;

[0049] eliminating outliers;

[0050] calculating a weighted average of the remaining measurement values as a final volume estimate.

[0051] Optionally, the image segmentation module adopts an improved U-Net model as a deep learning segmentation model, adopts ResNet18 as an encoder of the U-Net model, and introduces a batch normalization layer in each residual block of the ResNet18; and a spatial attention mechanism is introduced in each encoding block and decoding block of the U-Net model.

[0052] Optionally, the system further comprises a report generation module configured to generate a diagnostic report containing the prostate volume measurement result.

[0053] The application provides a prostate volume intelligent measurement device based on multi-modal images, comprising a memory and a processor.

[0054] The memory is configured to store a computer program.

[0055] The processor is configured to implement the multi-modal image-based intelligent prostate volume measurement method according to any one of the preceding embodiments when executing the computer program.

[0056] The application provides a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, the multi-modal image-based intelligent prostate volume measurement method according to any one of the preceding embodiments is implemented.

[0057] The application has the following advantages:

[0058] Firstly, by multi-modal image fusion, the application makes full use of the real-time nature of ultrasound imaging and the high-resolution characteristics of magnetic resonance imaging, effectively overcoming the limitations of single modality imaging. This fusion not only improves the accuracy of prostate boundary recognition, but also enhances the adaptability of the method to different imaging conditions.

[0059] Secondly, the application uses an improved deep learning model, especially introduces a spatial attention mechanism and a multi-scale feature extraction technology, which significantly improves the accuracy of prostate region segmentation. This improvement enables the model to better handle individual differences in prostate shape and size, thereby improving the accuracy of volume measurement.

[0060] Thirdly, the application innovatively introduces an iterative optimization and outlier processing mechanism, which effectively reduces the influence of random errors through multiple measurements and statistical analysis. This not only improves the reliability of the measurement results, but also provides more comprehensive data support for clinicians.

[0061] In addition, the method of the application realizes the full-flow automation from image acquisition to diagnostic report generation. This greatly improves work efficiency, reduces errors caused by human operation, and the standardized report format helps doctors quickly and accurately interpret the results.

[0062] Finally, the modular design and flexible parameter adjustment mechanism of the system make it have good scalability and adaptability. This means that the method can be easily applied to different medical devices and clinical environments, and can be optimized and upgraded according to specific needs.

[0063] In summary, the multi-modal image-based intelligent prostate volume measurement method and system provided by the application not only solves the problems of accuracy, efficiency and reliability in the prior art, but also provides more comprehensive and reliable data support for the diagnosis and treatment of prostate diseases. This innovative solution is expected to significantly improve the diagnosis level and treatment effect of prostate diseases, bringing blessings to the majority of patients. BRIEF DESCRIPTION OF DRAWINGS

[0064] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following will briefly introduce the drawings needed to be used in the embodiments description. Obviously, the drawings in the following description only some embodiments of the present application, and for those skilled in the art, other drawings can also be obtained without creative labor based on these drawings.

[0065] Figure 1 The flow chart of the prostate volume intelligent measurement method based on multi-modal images of the present application.

[0066] Figure 2 The structural diagram of the prostate volume intelligent measurement system based on multi-modal images of the present application.

[0067] Figure 3 The working flow chart of the image registration module of the present application.

[0068] Figure 4 The working flow chart of the measurement optimization module of the present application. DETAILED DESCRIPTION

[0069] In order to make the purpose, technical solutions and advantages of the present application more clear, the following will further describe the embodiments of the present application in combination with the drawings.

[0070] Embodiment one:

[0071] The present embodiment provides a prostate volume intelligent measurement method based on multi-modal images, comprising:

[0072] Step 1: obtaining the ultrasound image and the nuclear magnetic resonance image of the prostate region of the subject to be measured and pre-processing;

[0073] Step 2: registering the pre-processed ultrasound image and the nuclear magnetic resonance image, and the registration process comprises:

[0074] Step 21: constructing an adaptive similarity measurement function, calculating the similarity of the ultrasound image and the nuclear magnetic resonance image, and the calculation method is:

[0075] S = a · MI (I1, I2) + (1-a) · SSIM (I1, I2)

[0076] Wherein, MI represents mutual information, SSIM represents structural similarity index, a is the weight coefficient, and I1 and I2 represent two images to be registered respectively;

[0077] Step 22: taking the adaptive similarity measurement function as the objective function, and searching for the optimal registration parameter by using the particle swarm optimization algorithm;

[0078] Step 23: based on the optimal registration parameter, the ultrasound image and the nuclear magnetic resonance image are spatially aligned;

[0079] Step 3: Based on the registration results, fuse the ultrasound image and the MRI image to generate an enhanced image;

[0080] Step 4: Based on the enhanced image, segment the prostate region using a deep learning model;

[0081] Step 5: Based on the segmentation results, calculate the prostate volume, perform multiple measurements to obtain a volume measurement sequence, and perform statistical analysis on the volume measurement sequence to obtain an estimated prostate volume.

[0082] Example 2:

[0083] This embodiment provides an intelligent prostate volume measurement method based on multimodal imaging. By fusing ultrasound and MRI images, it achieves high-precision measurement of prostate volume. The method includes:

[0084] Step 1: Obtain ultrasound and MRI images of the prostate region of the subject.

[0085] Both imaging modalities have their advantages: ultrasound imaging is characterized by its real-time performance and low cost, while MRI imaging provides higher soft tissue contrast. By combining these two modalities, their complementary advantages can be fully utilized.

[0086] Step 2: Preprocess the ultrasound and MRI images.

[0087] For MRI images, grayscale conversion is performed first, transforming the color image into a grayscale image. This helps reduce data complexity and improve processing efficiency. Preferably, grayscale conversion can be performed using a weighted average method, where the weights for the red, green, and blue channels can be set to 0.299, 0.587, and 0.114, respectively. These weight values ​​are determined based on the human eye's sensitivity to different colors.

[0088] The MRI images are then denoised. Denoising is a crucial step in improving image quality and can be achieved using various algorithms, such as Gaussian filtering, median filtering, or wavelet transform. This embodiment uses an adaptive median filtering algorithm, which effectively removes salt-and-pepper noise while preserving image details. The algorithm's window size can be dynamically adjusted according to the noise level, typically between 3x3 and 7x7.

[0089] Normalization is the final step in preprocessing; it scales the image pixel values ​​to a specific range, typically [0,1] or [-1,1]. This helps improve the comparability between different images and is beneficial for training deep learning models. Normalization can be performed using the following formula:

[0090]

[0091] Where I is the original pixel value, I min and I max These are the minimum and maximum pixel values ​​of the image, respectively.

[0092] For ultrasound image preprocessing, denoising is also performed. However, considering the characteristics of ultrasound images, different denoising algorithms may be required. For example, a nonlocal mean denoising algorithm can be used, which performs excellently in preserving image edge and texture information. Furthermore, this embodiment also enhances the contrast of the ultrasound images to improve their recognizability. Histogram equalization or adaptive histogram equalization techniques can be used. In a preferred embodiment, a contrast-limited adaptive histogram equalization (CLAHE) algorithm can be used, which effectively enhances local contrast while avoiding excessive noise amplification.

[0093] Step 3: Register the preprocessed ultrasound and MRI images.

[0094] Step 31: Construct an adaptive similarity metric function to calculate the similarity between ultrasound images and MRI images.

[0095] An adaptive similarity metric function is crucial for achieving accurate registration. Considering the different characteristics of ultrasound and MRI images, this embodiment employs a hybrid metric function that combines mutual information and structural similarity. This function can be expressed as:

[0096] S=α·MI(I1,I2)+(1-α)·SSIM(I1,I2)

[0097] Where MI represents mutual information, SSIM represents the structural similarity index, α is the weighting coefficient, and I1 and I2 represent the two images to be registered. The weighting coefficient α can be dynamically adjusted according to the specific application scenario, and its value is usually in the range of [0.4, 0.6]. In a preferred embodiment, α can be set to 0.5 to balance the contributions of mutual information and structural similarity.

[0098] The similarity between ultrasound images and MRI images is calculated based on this similarity metric function. During the calculation, a multi-resolution strategy can be employed, gradually transitioning from low resolution to high resolution, which helps improve registration efficiency and robustness.

[0099] Step 32: Determine the optimal registration parameters.

[0100] An improved particle swarm optimization algorithm is employed, which efficiently searches for the optimal solution by simulating swarm intelligence behavior. The objective function of the algorithm is the aforementioned similarity metric function. The particle positions represent registration parameters, including translation, rotation, and scaling. The iterative process of the algorithm can be represented as follows:

[0101]

[0102] Among them, among them, and Let represent the velocity and position of the i-th particle at the t-th iteration, respectively; w be the inertial weight; c1 and c2 be the acceleration constants; and r1 and r2 be random numbers between [0,1]. For the individual optimal position of the particle, g t The optimal global position is determined. In this embodiment, the number of particles is set to 50, and the maximum number of iterations is 100. The inertia weight w can decrease linearly, with an initial value of 0.9 and a final value of 0.4. The acceleration constants c1 and c2 can both be set to 2.0. These parameter settings can achieve good convergence performance in most cases.

[0103] Step 33: Based on the optimized registration parameters, spatially align the ultrasound images and MRI images.

[0104] Alignment typically involves image interpolation, which can be achieved using bicubic interpolation, a method that strikes a good balance between maintaining image quality and computational efficiency.

[0105] Through the steps described above, this embodiment achieves precise registration of ultrasound and MRI images, laying the foundation for subsequent image fusion and segmentation. This multimodal registration method not only improves the accuracy of prostate region identification but also helps to fully utilize the complementary information of the two imaging modalities, thereby enhancing the final volume measurement accuracy.

[0106] Step 4: Based on the registration results, fuse the ultrasound image and the MRI image to generate an enhanced image.

[0107] Step 5: Based on the enhanced image, perform prostate region segmentation using a deep learning model.

[0108] This embodiment employs an improved U-Net model as the deep learning segmentation model. The U-Net model is renowned for its excellent image segmentation performance, and this invention further improves upon it to meet the specific requirements of prostate volume measurement.

[0109] First, this embodiment selects ResNet18 as the encoder of U-Net. ResNet18 is a deep residual network, and its unique skip connection structure can effectively alleviate the gradient vanishing problem of deep networks, thereby allowing the network to be further deepened and extract richer features. In this embodiment, each convolutional layer of ResNet18 is used as the downsampling path of U-Net, a design that significantly improves the model's feature extraction capability.

[0110] Preferably, this embodiment introduces a batch normalization layer in each residual block of ResNet18. Batch normalization can accelerate network convergence and improve the model's generalization ability. The mathematical expression for batch normalization is as follows:

[0111]

[0112] Where x is the input, E[x] and Var[x] are the mean and variance of the mini-batch, respectively, and ∈ is a very small constant (usually 10). -5 To prevent division by zero, γ and β are learnable scaling and translation parameters.

[0113] Secondly, this embodiment introduces a spatial attention mechanism into the U-Net model. Spatial attention helps the model focus on important regions in an image, which is particularly effective for organs like the prostate, whose shape and location are relatively fixed. The spatial attention module used in this embodiment can be represented as:

[0114] M(F)=σ(f 7×7 ([AvgPool(F);MaxPool(F)]))

[0115] Where F is the input feature map, f 7×7 This represents a 7x7 convolution operation, where AvgPool and MaxPool represent average pooling and max pooling, respectively, and σ is the sigmoid activation function. This attention module is inserted into each encoding and decoding block of U-Net, significantly improving the model's ability to perceive the prostate region.

[0116] Finally, this embodiment uses a cross-modal loss function for model training. Considering the different characteristics of ultrasound images and MRI images, a single loss function may be insufficient to cover the characteristics of both modalities. Therefore, this embodiment designs a hybrid loss function:

[0117] L=αL dice +βL focal +γL boundary

[0118] Among them, L dice It is the Dice loss, used to measure the overall segmentation accuracy; L focal It uses Focal loss, which is better at handling class imbalance problems; L boundaryThis is the boundary loss, used to improve the accuracy of segmentation boundaries. α, β, and γ are weighting coefficients used to balance the contributions of each loss term. In this embodiment, α = 0.5, β = 0.3, and γ = 0.2. This improved U-Net model significantly improves the accuracy of prostate region segmentation, laying a solid foundation for subsequent volumetric calculations.

[0119] The prostate region segmentation process of this invention includes multiple steps, aiming to achieve high-precision segmentation results. First, this method performs multi-scale feature extraction on the enhanced image. Multi-scale feature extraction can capture image information at different scales, which is beneficial for processing prostates of different sizes and shapes. In one embodiment of this invention, dilated convolution can be used to achieve multi-scale feature extraction. The mathematical expression for dilated convolution is as follows:

[0120]

[0121] Where F is the input feature map, k is the convolution kernel, and l is the dilation rate. By setting different dilation rates, the receptive field can be expanded without increasing the number of parameters, thereby capturing multi-scale information.

[0122] Next, this embodiment performs pixel-level classification based on the extracted features. This step uses a fully convolutional network (FCN) structure to map the feature map onto a segmentation map of the same size as the input image. During classification, the softmax function can be used to convert the feature vector of each pixel into a class probability distribution:

[0123]

[0124] Among them, y i x is the predicted category of pixel i. i is the feature vector of that pixel, and j and k are the class indices.

[0125] Then, this embodiment utilizes morphological prior constraints to optimize the classification results. The prostate gland has relatively fixed morphological features, and this prior knowledge can effectively improve segmentation accuracy. In a preferred embodiment, morphological opening and closing operations can be used to remove small misclassified regions and fill small holes. These operations can be represented as:

[0126]

[0127] Where A is the segmentation result and B is the structuring element. and These represent corrosion and expansion operations, respectively.

[0128] Finally, this embodiment performs edge trimming and boundary fusion to obtain the final segmentation result. Edge trimming aims to refine the segmentation boundary and can be implemented using an Active Contour Model. The Active Contour Model adjusts the segmentation boundary by minimizing an energy function:

[0129] E=∫0 1 [α|C′(s)| 2 +β|C″(s)| 2 +E ext (C(s))]ds

[0130] Where C(s) is the parameterized profile curve, α and β are parameters controlling the smoothness of the curve, and E ext It is an external energy term that reflects image features.

[0131] Boundary fusion integrates segmentation results from different scales or modalities. Weighted averaging or voting mechanisms can be used to fuse multiple segmentation results to obtain a more robust final segmentation.

[0132] Through this series of meticulously designed segmentation steps, the present invention can achieve high-precision segmentation of the prostate region, providing a reliable foundation for subsequent volume calculation.

[0133] Furthermore, this embodiment also includes a data augmentation step, which is a key technique for improving the performance of deep learning models. Data augmentation can expand the size and diversity of the training dataset, thereby improving the model's generalization ability and robustness.

[0134] First, this embodiment determines the region of interest (ROI) of the prostate region based on the segmentation results. This step aims to focus on the most relevant image region and avoid introducing unnecessary background noise.

[0135] In this embodiment, the region of interest is defined as the minimum bounding rectangle of the prostate contour, and appropriately expanded by a certain proportion (e.g., 10%) in each direction to include more contextual information. The region of interest can be determined through the following steps:

[0136] 1. Calculate the bounding box of the prostate segmentation mask.

[0137] 2. Expand the bounding box by 10% in each direction.

[0138] 3. Crop the original image to obtain the region of interest.

[0139] Next, this method performs transformations such as rotation, scaling, and translation on the region of interest. These geometric transformations can simulate different imaging angles and scales, increasing the model's adaptability to various situations. In the rotation transformation, the rotation angle can be randomly selected within the range of [-15°, 15°]. The rotation transformation can be implemented using the following matrix:

[0140]

[0141] Where θ is the rotation angle.

[0142] For scaling transformations, a scaling factor can be randomly selected within the range [0.9, 1.1]. The scaling transformation can be implemented using the following matrix:

[0143]

[0144] Among them, s x and s y These are the scaling factors in the x and y directions, respectively.

[0145] Translation transformations can be performed by randomly selecting a translation distance within ±5% of the image size. Translation transformations can be implemented using the following matrix:

[0146]

[0147] Among them, t x and t y These are the translation distances in the x and y directions, respectively.

[0148] Through these transformations, this method generates enhanced training samples. It is important to note that these transformations should be applied to both the image and the corresponding segmentation mask to maintain consistency.

[0149] Finally, the augmented training samples are added to the training dataset. Preferably, the size of the original dataset can be expanded by 2 to 3 times. For example, if the original dataset contains 1,000 images, the training dataset may contain 2,000 to 3,000 images after data augmentation. This expansion can significantly improve the model's learning ability and generalization performance.

[0150] Through various data augmentation steps, this invention can effectively solve the common problem of data scarcity in the field of medical imaging, improve the performance and reliability of deep learning models, and thus provide a more solid foundation for the accurate measurement of prostate volume.

[0151] Step 6: Calculate the prostate volume based on the segmentation results.

[0152] Step 61: Perform 3D reconstruction on the segmentation results.

[0153] This embodiment uses the Marching Cubes Algorithm to achieve 3D reconstruction. The basic idea of ​​this algorithm is to traverse each cell in the voxel mesh and generate a triangular mesh based on the intersection of isosurfaces and cell vertices. The key steps of the algorithm can be represented as follows:

[0154] (x,y,z)=f(x,y,z)isovalue

[0155] Here, f(x,y,z) is the value of the voxel, and isovalue is the isosurface threshold. When the sign of V(x,y,z) changes among the eight vertices of the cell, it indicates that the isosurface passes through the cell, and the corresponding triangular facet needs to be generated.

[0156] Step 62: Calculate the prostate volume based on the reconstruction results.

[0157] Volume calculation can be achieved by voxelizing the 3D model and then counting the number of voxels.

[0158] Another more precise method is to use the Gaussian divergence theorem to transform the volume calculation into a surface integral:

[0159]

[0160] Where r is the vector from a point on the surface to the origin, n is the surface normal vector, and S is the prostate surface.

[0161] Step 63: Perform multiple measurements to obtain a volume measurement sequence.

[0162] To further improve the reliability of the measurement, this embodiment performs multiple measurements to obtain a volume measurement sequence. Multiple measurements help reduce the impact of random errors. Preferably, 5 to 10 measurements can be performed, which typically achieves a good balance between accuracy and efficiency.

[0163] Step 64: Perform statistical analysis on the volume measurement sequence to obtain the final volume estimate.

[0164] Statistical analysis can include outlier detection and central tendency estimation.

[0165] In outlier detection, this embodiment uses an improved Z-score method:

[0166]

[0167] Where, x i Z is a single measurement, X is the measurement sequence, and MAD is the median absolute deviation. i If the value exceeds a certain threshold (e.g., 3.5), the measurement is considered an outlier.

[0168] In central tendency estimation, considering the possibility of non-normal distribution, the median or truncated mean can be used as the final volume estimate. The formula for calculating the truncated mean is:

[0169]

[0170] Where, x (i) These are the sorted measurements, and k is the truncation count, which is usually 10% of the number of measurements.

[0171] By employing this multi-step, multiple-measurement method, this embodiment can significantly improve the accuracy and reliability of prostate volume measurement, providing strong support for clinical diagnosis and treatment decisions.

[0172] In addition, this embodiment also employs the following steps to optimize volume measurement.

[0173] (1) Set the volume measurement threshold and the maximum number of measurements.

[0174] The measurement threshold is used to determine whether the measurement result has reached sufficient accuracy, while the maximum number of measurements is used to prevent time wastage due to over-measurement. In this embodiment, the volume measurement threshold is set to 1% of the previous measurement result, and the maximum number of measurements is set to 15. The selection of these parameters is based on extensive clinical experience and achieves a good balance between accuracy and efficiency.

[0175] (2) Calculate the difference between two consecutive measurement results.

[0176] The difference is calculated using relative error, and the formula is as follows:

[0177]

[0178] Among them, V i This is the current measured value, V i-1 This is the value from the previous measurement.

[0179] (3) When the difference is less than the threshold or the maximum number of measurements is reached, the measurement is stopped.

[0180] (4) Perform outlier detection and removal based on the measurement sequence.

[0181] Outliers can be caused by various factors, such as image noise and segmentation errors. To identify outliers, a modified Z-score method can be used:

[0182]

[0183] Where, x i Z is a single measurement, X is the measurement sequence, and MAD is the median absolute deviation.i If the value exceeds a certain threshold (usually 3), the measurement is considered an outlier and is removed.

[0184] (5) Calculate the weighted average of the remaining measurements as the final volume estimate.

[0185] Weighted averages can account for the reliability of different measurement results, thus yielding a more accurate estimate. Weights can be determined based on position within the measurement sequence, with newer measurements typically assigned higher weights. The formula for calculating a weighted average is as follows:

[0186]

[0187] Among them, V i It is the i-th valid measurement value, w i These are the corresponding weights. This example uses exponentially decaying weights: w i =e -λ(n-i) , where λ is the attenuation rate (e.g., 0.1) and n is the number of effective measurements.

[0188] Step 7: Generate a diagnostic report that includes the results of prostate volume measurement.

[0189] Step 71: Extract prostate volume measurement results and historical measurement data.

[0190] This includes not only current measurements but also the patient's past measurement records (if any). Historical patient data can be retrieved from the electronic health record system and integrated with current measurements. Data extraction can be performed using Structured Query Language (SQL) or other appropriate database query methods.

[0191] Step 72: Calculate the volume change trend.

[0192] Volume change trends are crucial for assessing disease progression or treatment effectiveness, and volume change trends can be calculated in the following ways:

[0193] 1. Absolute change: ΔV = V current -V previous ;

[0194] 2. Relative change:

[0195] 3. Annualized growth rate:

[0196] Among them, V current V is the currently measured volume. previous This is the volume measured last time.

[0197] Step 93: Generate preliminary diagnostic suggestions based on preset diagnostic rules.

[0198] Diagnostic criteria can be developed based on the absolute value of prostate volume, rate of change, and other relevant clinical indicators. For example, the following criteria may be used:

[0199] If the prostate volume is <30mL, it is considered to be of normal size.

[0200] If the prostate volume is 30mL ≤ prostate volume < 50mL, it is considered to be mildly enlarged.

[0201] If the prostate volume is 50mL ≤ prostate volume < 80mL, it is considered moderately enlarged.

[0202] If the prostate volume is ≥80mL, it is considered severely enlarged.

[0203] Additionally, suggestions can be provided based on the rate of change of volume:

[0204] If the annual growth rate is less than 5%, routine follow-up is recommended.

[0205] If the annual growth rate is less than 10% and less than 5%, it is recommended to shorten the follow-up interval.

[0206] If the annual growth rate is ≥10%, further clinical evaluation and possible treatment interventions are recommended.

[0207] Step 94: Integrate measurement results, trends, and diagnostic recommendations to generate a standardized diagnostic report.

[0208] Finally, this embodiment integrates measurement results, trends, and diagnostic recommendations to generate a standardized diagnostic report. The report should include, but is not limited to:

[0209] 1. Patient's basic information;

[0210] 2. Check the date and method;

[0211] 3. Current prostate volume measurement results;

[0212] 4. Comparison with historical measurement data (if applicable);

[0213] 5. Analysis of volume change trends;

[0214] 6. Preliminary diagnosis based on volume;

[0215] 7. Doctor's recommendations and follow-up plan;

[0216] Preferably, the diagnostic report can be in a structured format for easy reading and understanding by doctors. Additionally, visual elements, such as volume change trend charts, can be incorporated to visually demonstrate changes in the patient's condition.

[0217] Through this comprehensive diagnostic report generation process, the present invention not only provides accurate prostate volume measurement results, but also provides clinicians with valuable diagnostic references and treatment suggestions, thereby improving the diagnosis and management of prostate diseases.

[0218] Example 3:

[0219] This embodiment provides an intelligent prostate volume measurement system based on multimodal imaging, including:

[0220] The image acquisition module is used to acquire ultrasound and MRI images of the patient's prostate region.

[0221] The image preprocessing module is used to preprocess ultrasound images and MRI images;

[0222] The image registration module is used to perform spatial alignment of ultrasound images and MRI images;

[0223] The registration process includes:

[0224] An adaptive similarity metric function is constructed to calculate the similarity between ultrasound images and MRI images. The calculation method is as follows:

[0225] S=α·MI(I1,I2)+(1-α)·SSIM(I1,I2)

[0226] Where MI represents mutual information, SSIM represents structural similarity index, α is the weight coefficient, and I1 and I2 represent the two images to be registered, respectively.

[0227] Using an adaptive similarity metric function as the objective function, the particle swarm optimization algorithm is employed to search for the optimal registration parameters.

[0228] Based on the optimal registration parameters, the ultrasound images and MRI images are spatially aligned.

[0229] The image fusion module is used to fuse registered ultrasound images and MRI images to generate enhanced images;

[0230] The image segmentation module is used to perform prostate region segmentation on the enhanced image;

[0231] The volume calculation module is used to calculate the prostate volume based on the segmentation results, perform multiple measurements to obtain a volume measurement sequence, and perform statistical analysis on the volume measurement sequence to obtain an estimated prostate volume value.

[0232] Some steps in the embodiments of the present invention can be implemented using software, and the corresponding software program can be stored in a readable storage medium, such as an optical disc or a hard disk.

[0233] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for intelligent measurement of prostate volume based on multimodal imaging, characterized in that, The method includes: Step 1: Acquire ultrasound and MRI images of the prostate region of the subject and perform preprocessing; Step 2: Register the preprocessed ultrasound and MRI images. The registration process includes: Step 21: Construct an adaptive similarity metric function to calculate the similarity between ultrasound images and MRI images. The calculation method is as follows: S=α·MI(I1,I2)+(1-α)·SSIM(I1,I2) Where MI represents mutual information, SSIM represents structural similarity index, α is the weight coefficient, and I1 and I2 represent the two images to be registered, respectively. Step 22: Using the adaptive similarity metric function as the objective function, the particle swarm optimization algorithm is used to search for the optimal registration parameters; Step 23: Based on the optimal registration parameters, spatially align the ultrasound image and the MRI image; Step 3: Based on the registration results, fuse the ultrasound image and the MRI image to generate an enhanced image; Step 4: Based on the enhanced image, segment the prostate region using a deep learning model; Step 5: Calculate the prostate volume based on the segmentation results, perform multiple measurements to obtain a volume measurement sequence, and perform statistical analysis on the volume measurement sequence to obtain an estimated prostate volume.

2. The method according to claim 1, characterized in that, Step 5 also includes volume measurement optimization, the optimization process of which includes: Step 51: Set the volume measurement error threshold and the maximum number of measurements; Step 52: Calculate the relative error between two adjacent measurements in the volume measurement sequence. The calculation formula is as follows: Among them, V i This is the current measured value, V i-1 This is the previous measurement value; Step 53: Stop the measurement when the relative error is less than the volume measurement error threshold or when the maximum number of measurements is reached; Step 54: Remove outliers; Step 55: Calculate the weighted average of the remaining measurements as the final volume estimate.

3. The method according to claim 1, characterized in that, Step 4 employs an improved U-Net model as the deep learning segmentation model, uses ResNet18 as the encoder of the U-Net model, and introduces a batch normalization layer in each residual block of ResNet18; a spatial attention mechanism is introduced in each encoding block and decoding block of the U-Net model.

4. The method according to claim 1, characterized in that, The method uses a cross-modal loss function to train the deep learning model, and the cross-modal loss function is: L=αL dice +βL focal +γL boundary Among them, L dice It is Dice's loss, L focal It is Focal loss, L boundary It is the boundary loss, used to improve the accuracy of the segmentation boundary. α, β and γ are weighting coefficients.

5. A prostate volume intelligent measurement system based on multimodal imaging, characterized in that, The system includes: The image acquisition module is used to acquire ultrasound and MRI images of the patient's prostate region. An image preprocessing module is used to preprocess the ultrasound images and MRI images; The image registration module is used to perform spatial alignment of the ultrasound image and the MRI image; The registration process includes: An adaptive similarity metric function is constructed to calculate the similarity between ultrasound images and MRI images. The calculation method is as follows: S=α·MI(I1,I2)+(1-α)·SSIM(I1,I2) Where MI represents mutual information, SSIM represents structural similarity index, α is the weight coefficient, and I1 and I2 represent the two images to be registered, respectively. Using the adaptive similarity metric function as the objective function, the particle swarm optimization algorithm is employed to search for the optimal registration parameters; Based on the optimal registration parameters, the ultrasound images and MRI images are spatially aligned; The image fusion module is used to fuse registered ultrasound images and MRI images to generate enhanced images; An image segmentation module is used to perform prostate region segmentation on the enhanced image; The volume calculation module is used to calculate the prostate volume based on the segmentation results, perform multiple measurements to obtain a volume measurement sequence, and perform statistical analysis on the volume measurement sequence to obtain an estimated prostate volume value.

6. The system according to claim 5, characterized in that, The system also includes a measurement optimization module, the optimization process of which includes: Set the volume measurement error threshold and the maximum number of measurements; The relative error between two adjacent measurements in the volume measurement sequence is calculated using the following formula: Among them, V i This is the current measured value, V i-1 This is the previous measurement value; Measurement is stopped when the relative error is less than the volume measurement error threshold or when the maximum number of measurements is reached; Remove outliers; Calculate the weighted average of the remaining measurements as the final volume estimate.

7. The system according to claim 5, characterized in that, The image segmentation module uses an improved U-Net model as the deep learning segmentation model, employs ResNet18 as the encoder of the U-Net model, and introduces a batch normalization layer in each residual block of the ResNet18 model; a spatial attention mechanism is introduced in each encoding block and decoding block of the U-Net model.

8. The system according to claim 5, characterized in that, The system also includes a report generation module for generating a diagnostic report containing prostate volume measurement results.

9. A smart device for measuring prostate volume based on multimodal imaging, characterized in that, Including memory and processor; The memory is used to store computer programs; The processor is configured to, when executing the computer program, implement the intelligent prostate volume measurement method based on multimodal imaging as described in any one of claims 1 to 4.

10. A computer-readable storage medium, characterized in that, The storage medium stores a computer program, which, when executed by a processor, implements the intelligent prostate volume measurement method based on multimodal imaging as described in any one of claims 1 to 4.

Images