Medical image quality evaluation method, device and computer equipment

Abstract

The application relates to a medical image quality evaluation method and device and a computer device. First, second and third statistical parameters corresponding to pixel values of all pixels in a foreground image of a medical image to be evaluated are obtained; for each target image block, an active state of the target image block is determined according to the first statistical parameter and pixel values corresponding to the target image block; a noise state of the target image block is determined according to the second statistical parameter and the pixel values corresponding to the target image block; a damaged state of the target image block is determined according to the third statistical parameter and the pixel values corresponding to the target image block; and a quality evaluation result of the medical image to be evaluated is determined according to the active state, the noise state and the damaged state of each target image block. The embodiments of the application consider three quality indexes of the active state, the noise state and the damaged state, and improve the objectivity, the robustness and the clinical relevance of the quality evaluation.

Description

Technical Field

[0001] This application relates to the field of medical technology, and in particular to a method, apparatus and computer device for medical image quality assessment. Background Technology

[0002] Positron emission tomography (PET) reconstructs the distribution of radiopharmaceuticals in the body by detecting 511 keV gamma photons generated by positron-electron annihilation, thereby reflecting tissue metabolic levels, blood flow changes, and cellular functional states. Therefore, PET with higher image quality has irreplaceable value in early tumor diagnosis, monitoring of neurological diseases, assessment of myocardial perfusion, and pharmacokinetics studies of novel tracers.

[0003] Currently, algorithms such as Blind / Referenceless Image Spatial Quality Evaluator (BRISQUE), Natural Image Quality Evaluator (NIQE), and Perception-based Image Quality Evaluator (PIQE) are mainly used to evaluate the quality of PET images.

[0004] However, the image quality assessment methods described above are inaccurate for assessing the quality of PET images. Summary of the Invention

[0005] Therefore, it is necessary to provide a medical image quality assessment method, apparatus, and computer equipment that can improve the accuracy of PET image quality assessment in response to the above-mentioned technical problems.

[0006] In a first aspect, this application provides a method for medical image quality assessment, including:

[0007] Obtain a first statistical parameter, a second statistical parameter, and a third statistical parameter corresponding to the pixel values ​​of all pixels in the foreground image of the medical image to be evaluated; the first statistical parameter includes the mean, range, and preset percentile of all pixel values, the second statistical parameter includes the coefficient of variation, and the third statistical parameter includes the signal-to-noise ratio;

[0008] For each target image block, the active state of the target image block is determined based on the first statistical parameter and the pixel value corresponding to the target image block; the target image block is obtained by dividing the foreground image.

[0009] The noise state of the target image block is determined based on the second statistical parameter and the pixel value corresponding to the target image block;

[0010] The damage state of the target image block is determined based on the third statistical parameter and the pixel value corresponding to the target image block;

[0011] The quality assessment result of the medical image to be evaluated is determined based on the active state, noise state, and damaged state of each target image block.

[0012] In one embodiment, determining the activity state of the target image patch based on the first statistical parameter and the pixel value corresponding to the target image patch includes:

[0013] The intensity and foreground factor of the target image block are determined based on the pixel values ​​corresponding to the target image block.

[0014] Based on the relationship between the mean and the preset percentile, a candidate intensity threshold is determined, and the candidate intensity threshold is compensated based on the range to obtain the intensity threshold.

[0015] When the intensity is greater than the intensity threshold and the foreground factor is greater than the preset foreground factor threshold, the active state is determined to characterize the target image block as an active image block.

[0016] In one embodiment, determining the noise state of the target image block based on the second statistical parameter and the pixel value corresponding to the target image block includes:

[0017] Based on the pixel values ​​corresponding to the target image patch, determine the expected standard deviation and observed standard deviation of the target image patch;

[0018] The excess noise ratio and Poisson perception bias of the target image patch are determined based on the expected standard deviation and the observed standard deviation.

[0019] The noise baseline of the target image patch is determined based on the excess noise ratio, the Poisson perception bias, and the second statistical parameter.

[0020] The noise state of the target image patch is determined based on the observation standard deviation and the noise benchmark.

[0021] In one embodiment, determining the noise state of the target image patch based on the observation standard deviation and the noise benchmark includes:

[0022] Determine the product of the noise reference and the preset noise value;

[0023] If the observed standard deviation is greater than the product result, the noise state of the target image block is determined to characterize the target image block as a noisy image block.

[0024] In one embodiment, determining the noise baseline of the target image patch based on the excess noise ratio, the Poisson perceptual bias, and the second statistical parameter includes:

[0025] If the coefficient of variation is greater than a preset threshold, the noise benchmark is determined based on the first weighting coefficient, the excess noise ratio, and the Poisson perception bias.

[0026] If the coefficient of variation is not greater than the preset threshold, the noise benchmark is determined based on the second weighting coefficient, the third weighting coefficient, the excess noise ratio, and the Poisson perception bias.

[0027] In one embodiment, determining the damage state of the target image patch based on the third statistical parameter and the pixel value corresponding to the target image patch includes:

[0028] The target image block is divided into multiple line segments by dividing each border.

[0029] For each line segment, the segment standard deviation corresponding to the line segment is determined based on the pixel values ​​on the line segment;

[0030] The threshold for damaged blocks is determined based on the third statistical parameter.

[0031] If at least one of the segment standard deviations is greater than or equal to the damaged block threshold, the damaged state characterizes the target image block as a damaged image block.

[0032] In one embodiment, determining the quality assessment result of the medical image to be evaluated based on the activity state, noise state, and damaged state of each target image patch includes:

[0033] When the noise state characterizes the target image block as a noisy image block, the variance corresponding to the pixel values ​​of all pixels in the target image block is determined as the noise quantization value of the target image block;

[0034] When the damaged state characterizes the target image block as a damaged image block, the maximum variance corresponding to the pixel values ​​of all pixels on each line segment of the target image block is determined as the damage quantization value of the target image block;

[0035] When the active state indicates that the target image block is an active image block, the evaluation value of the target image block is determined based on the noise quantization value and / or the damaged quantization value.

[0036] The quality assessment result is determined based on the evaluation value of each target image patch.

[0037] In one embodiment, determining the quality assessment result based on the evaluation values ​​of each of the target image patches includes:

[0038] The target image block whose evaluation value is less than the preset evaluation value is identified as the image block to be evaluated;

[0039] The quality assessment result is determined based on the evaluation value of the image patch to be evaluated and the number of image patches to be evaluated.

[0040] In one embodiment, determining the quality assessment result based on the assessment value of the image patch to be evaluated and the number of image patches to be evaluated includes:

[0041] Determine the sum of the evaluation values ​​of each of the image blocks to be evaluated, and the first summation value of the preset constant value;

[0042] Determine a second summation value of the number of image patches to be evaluated and the preset constant value;

[0043] The quality assessment result is determined based on the ratio of the first summation value to the second summation value.

[0044] Secondly, this application also provides a medical image quality assessment device, comprising:

[0045] The acquisition module is used to acquire a first statistical parameter, a second statistical parameter, and a third statistical parameter corresponding to the pixel values ​​of all pixels in the foreground image of the medical image to be evaluated; the first statistical parameter includes the mean, range, and preset percentile of all pixel values, the second statistical parameter includes the coefficient of variation, and the third statistical parameter includes the signal-to-noise ratio;

[0046] The first determining module is used to determine the active state of each target image block based on the first statistical parameters and the pixel values ​​corresponding to the target image block; the target image block is obtained by dividing the foreground image.

[0047] The second determining module is used to determine the noise state of the target image block based on the second statistical parameters and the pixel values ​​corresponding to the target image block;

[0048] The third determining module is used to determine the damage state of the target image block based on the third statistical parameter and the pixel value corresponding to the target image block;

[0049] The fourth determining module is used to determine the quality assessment result of the medical image to be evaluated based on the active state, noise state, and damaged state of each target image block.

[0050] Thirdly, this application also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the method steps provided in the first aspect.

[0051] Fourthly, this application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the method steps provided in the first aspect.

[0052] Fifthly, this application also provides a computer program product, including a computer program that, when executed by a processor, implements the method steps provided in the first aspect.

[0053] The aforementioned medical image quality assessment method, apparatus, and computer equipment acquire first, second, and third statistical parameters corresponding to the pixel values ​​of all pixels in the foreground image of the medical image to be assessed; for each target image block, the activity state of the target image block is determined based on the first statistical parameter and the pixel values ​​corresponding to the target image block; the noise state of the target image block is determined based on the second statistical parameter and the pixel values ​​corresponding to the target image block; the damage state of the target image block is determined based on the third statistical parameter and the pixel values ​​corresponding to the target image block; and the quality assessment result of the medical image to be assessed is determined based on the activity state, noise state, and damage state of each target image block; the first statistical parameter includes the mean, range, and preset percentile of all pixel values; the second statistical parameter includes the coefficient of variation; and the third statistical parameter includes the signal-to-noise ratio; the target image block is obtained by dividing the foreground image. This application's embodiments extract multi-dimensional statistical parameters from the foreground image, and determine the active state, noise state, and damaged state by combining the pixel values ​​of the target image block. Finally, the multi-state results are fused to determine the quality assessment result of the medical image to be evaluated. This not only specifically avoids the interference of the background area of ​​the medical image on the quality assessment, but also realizes the hierarchical assessment from global statistical features to local block-level states, so that the quality assessment results are consistent with the physical laws of medical imaging. At the same time, it takes into account the three quality indicators of active state, noise state, and damaged state, solves the problem that traditional algorithms cannot adapt to the characteristics of medical images, improves the objectivity, robustness, and clinical relevance of the quality assessment, and can adapt to the quality assessment needs of medical images under different scanning conditions and different imaging devices. Attached Figure Description

[0054] To more clearly illustrate the technical solutions in the embodiments of this application or related technologies, the drawings used in the description of the embodiments of this application or related technologies will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0055] Figure 1 This is a diagram illustrating the application environment of a medical image quality assessment method in one embodiment.

[0056] Figure 2 This is a flowchart illustrating a medical image quality assessment method in one embodiment;

[0057] Figure 3 This is a flowchart illustrating an active state determination method in one embodiment;

[0058] Figure 4 This is a flowchart illustrating a noise state determination method in one embodiment;

[0059] Figure 5 This is a flowchart illustrating a method for determining the damaged state in one embodiment;

[0060] Figure 6 This is a flowchart illustrating a method for determining quality assessment results in one embodiment;

[0061] Figure 7 This is a flowchart illustrating a method for determining quality assessment results in another embodiment;

[0062] Figure 8 This is a first schematic diagram of a PET image in one embodiment;

[0063] Figure 9 This is a second schematic diagram of a PET image in one embodiment;

[0064] Figure 10 This is a third schematic diagram of a PET image in one embodiment;

[0065] Figure 11 This is a first bar chart of the quality assessment results in one embodiment;

[0066] Figure 12 This is a fourth schematic diagram of a PET image in one embodiment;

[0067] Figure 13 This is a fifth schematic diagram of a PET image in one embodiment;

[0068] Figure 14 This is a sixth schematic diagram of a PET image in one embodiment;

[0069] Figure 15 This is a second bar chart of the quality assessment results in one embodiment;

[0070] Figure 16 This is a seventh schematic diagram of a PET image in one embodiment;

[0071] Figure 17 This is an eighth schematic diagram of a PET image in one embodiment;

[0072] Figure 18 This is a ninth schematic diagram of a PET image in one embodiment;

[0073] Figure 19 This is a third bar diagram illustrating the quality assessment results in one embodiment;

[0074] Figure 20 This is a tenth schematic diagram of a PET image in one embodiment;

[0075] Figure 21 This is the eleventh schematic diagram of a PET image in one embodiment;

[0076] Figure 22 This is a twelfth schematic diagram of a PET image in one embodiment;

[0077] Figure 23 This is a schematic diagram of the fourth bar of the quality assessment results in one embodiment;

[0078] Figure 24 This is a structural block diagram of a medical image quality assessment device in one embodiment. Detailed Implementation

[0079] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0080] It should be noted that the terms "first," "second," etc., used in this application can be used to describe various elements, but these elements are not limited by these terms. These terms are only used to distinguish the first element from the second element. The terms "comprising" and "having," and any variations thereof, used in this application, are intended to cover non-exclusive inclusion. The term "multiple" used in this application refers to two or more. The term "and / or" used in this application refers to one of the embodiments, or any combination of multiple embodiments.

[0081] The medical image quality assessment method provided in this application embodiment can be applied to, for example... Figure 1The application environment shown includes a computer device, which may be a server, and its internal structure diagram may be as follows. Figure 1 As shown, this computer device includes a processor, memory, input / output interfaces (I / O), and a communication interface. The processor, memory, and I / O interfaces are connected via a system bus, and the communication interface is also connected to the system bus via the I / O interfaces. The processor provides computational and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and a database. The internal memory provides the environment for the operating system and computer programs stored in the non-volatile storage media. The database stores data related to quality assessment. The I / O interfaces are used for exchanging information between the processor and external devices. The communication interface is used for communicating with external terminals via a network. When the computer program is executed by the processor, it implements a medical image quality assessment method. The server can be a standalone physical server, a server cluster or distributed system consisting of multiple physical servers, or a cloud server providing cloud computing services.

[0082] Those skilled in the art will understand that Figure 1 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0083] In one exemplary embodiment, such as Figure 2 As shown, a medical image quality assessment method is provided, which can be applied to... Figure 1 The following explanation uses computer equipment as an example, including the following steps S201 to S205. Wherein:

[0084] S201, obtain the first statistical parameter, the second statistical parameter and the third statistical parameter corresponding to the pixel values ​​of all pixels in the foreground image of the medical image to be evaluated; the first statistical parameter includes the mean, range and preset percentile of all pixel values, the second statistical parameter includes the coefficient of variation and the third statistical parameter includes the signal-to-noise ratio.

[0085] Optionally, the medical image to be evaluated can be a PET image, a computed tomography (CT) image, an MRI image, etc. The medical image to be evaluated can be a two-dimensional image or a three-dimensional image.

[0086] Traditional PIQE algorithms divide the entire image into uniform blocks for analysis. However, the signal in the background region of medical images is close to zero, resulting in extremely low variance. This leads to background regions being misclassified as artifact blocks and low-count regions being misclassified as excessively noisy. The block-level weighting method is incompatible with the imaging characteristics of medical images, thus failing to provide accurate quality scores. Therefore, in this embodiment, Gaussian smoothing is used to process the medical image I to be evaluated, eliminating local noise (G = Gaussian(I)). The Sobel algorithm is then used to obtain a gradient map grad = Sobel(G). Based on the gradient map, a foreground seed (seeds = grad > preset value Tg) is determined. The watershed algorithm is then used to process the foreground seed and gradient map to obtain the foreground image fg = Watershed(grad, seeds). All subsequent evaluations are performed only within the foreground image.

[0087] The first statistical parameters include the mean, dynamic range, and preset percentiles of all pixel values, where the mean... The average value of all pixels can be expressed as: N is the number of pixels in the foreground image. The pixel value is the range value (dynamic range DR), which is the maximum pixel value. and minimum pixel value The difference between them can be specifically expressed as The preset percentiles include the fifth percentile and the tenth percentile. For example, if 90% of all pixel values ​​are greater than 20, then 20 is the tenth percentile.

[0088] The second statistical parameter includes the coefficient of variation (NL), which measures the relative fluctuation of pixel values ​​and is the ratio of the standard deviation to the mean. Specifically, it can be expressed as... , The standard deviation of the foreground image. To prevent constants with a denominator of 0.

[0089] The third statistical parameter includes the signal-to-noise ratio, which can be specifically expressed as: :

[0090] S202, for each target image block, determine the active state of the target image block based on the first statistical parameter and the pixel value corresponding to the target image block; the target image block is obtained by dividing the foreground image.

[0091] In this embodiment, dividing the foreground image into target image blocks can be achieved by directly dividing the foreground image into initial image blocks if the size of the foreground image is an integer multiple of the target image block size, and then normalizing the initial image blocks to obtain the target image blocks. If the size of the foreground image is not an integer multiple of the target image block size, the foreground image is padded, and the padded foreground image is then divided into initial image blocks. Optionally, the padded methods include: zero-value padded (preferred); mirror padded; and edge copy padded, thus ensuring that the size of each target image block is consistent.

[0092] In medical image evaluation, activity represents the degree of metabolic activity and is an important quality indicator; however, traditional PIQE or NIQE algorithms do not evaluate this factor at all. In this embodiment, the activity state of a target image block is determined based on a first statistical parameter and the pixel value corresponding to the target image block. In one possible implementation, an intensity threshold can be determined based on the first statistical parameter, and the intensity corresponding to the target image block can be determined based on the pixel value. If the intensity is greater than the intensity threshold, the activity state indicates that the target image block is an active image block.

[0093] In another possible implementation, an intensity threshold can be determined based on a first statistical parameter. Based on the pixel value corresponding to the target image block, the intensity and foreground factor corresponding to the target image block can be determined. If the intensity is greater than the intensity threshold and the foreground factor is greater than the foreground factor threshold, the active state indicator target image block is determined to be an active image block. The foreground factor is the proportion of the foreground in the target image block.

[0094] S203, determine the noise state of the target image block based on the second statistical parameter and the pixel value corresponding to the target image block.

[0095] In this embodiment, the expected standard deviation and observed standard deviation of the target image block can be determined based on the pixel values ​​corresponding to the target image block. Then, the excess noise ratio and Poisson perception bias of the target image block can be determined based on the expected standard deviation and observed standard deviation. Based on the excess noise ratio and Poisson perception bias, the noise benchmark of the target image block can be determined. Based on the observed standard deviation and the noise benchmark, the noise state of the target image block can be determined.

[0096] In another possible implementation, the pixel values ​​of the target image block are first divided into several subgroups according to intensity intervals. The pixel mean and variance of each subgroup are calculated. Then, the least squares method is used to perform a linear fit on the mean and variance of all subgroups, conforming to the characteristics of Poisson noise. After obtaining the fitted model, the fitting residuals of each subgroup are calculated, and the mean of the residuals is obtained. Further, the mean of the residuals is normalized by combining the coefficient of variation in the second statistical parameter. Finally, the normalized residual values ​​are compared with an adaptively adjusted residual threshold, and the noise state is determined based on the comparison results.

[0097] S204. Determine the damage status of the target image block based on the third statistical parameter and the pixel value corresponding to the target image block.

[0098] Lesions in medical images typically exhibit relatively uniform metabolic morphological features in a localized manner. However, conventional quality assessment algorithms rely solely on local variance and mean, failing to detect the unique metabolic structural features of medical images. Therefore, it is necessary to determine the damage state of the target image patch using a third statistical parameter and the pixel values ​​corresponding to the target image patch.

[0099] In this embodiment of the application, the borders of the target image block can be divided into multiple line segments. For each line segment, the segment standard deviation corresponding to the line segment is determined based on the pixel value on the line segment. The damaged block threshold is determined based on the third statistical parameter. If at least one segment standard deviation is greater than or equal to the damaged block threshold, the target image block representing the damaged state is determined to be a damaged image block.

[0100] In one possible implementation, the central region of the target image patch is used as the core, and the surrounding annular region is used as the comparison region. The pixel variance of the two regions is calculated separately, and the difference rate between the pixel variance of the central region and the surrounding region is obtained. The difference rate is then weighted and corrected based on the signal-to-noise ratio (SNR) in the third statistical parameter. If the weighted and corrected difference rate is greater than a threshold, it indicates that the metabolic structure of the center-surrounding part of the target image patch is abnormally distorted and does not conform to the uniform metabolic characteristics of a lesion. The target image patch is then determined to be a damaged image patch; otherwise, it is considered an undamaged image patch. A lower SNR results in a smaller weighting coefficient, reducing the probability of misjudgment in high-noise scenes.

[0101] S205. Based on the active state, noise state, and damaged state of each target image block, determine the quality assessment result of the medical image to be evaluated.

[0102] In this embodiment, when the noise state characterizes the target image block as a noisy image block, the variance corresponding to the pixel values ​​of all pixels in the target image block is determined as the noise quantization value of the target image block. When the damage state characterizes the target image block as a damaged image block, the maximum variance corresponding to the pixel values ​​of all pixels on each line segment of the target image block is determined as the damage quantization value of the target image block. If the active state of the target image block indicates that the target image block is an active image block, the evaluation value of the target image block is determined based on the noise quantization value and the damage quantization value, and the quality evaluation result is determined based on the evaluation values ​​of each target image.

[0103] If the active status of the target image block indicates that the target image block is not an active image block, regardless of whether the noise status indicates that the target image block is a noisy image block or the damaged status indicates that the target image block is a damaged image block, the evaluation value corresponding to the target image block can be directly determined to be 0, and the quality evaluation result can be determined based on the evaluation value of each target image block.

[0104] In one possible implementation, if the active state indicates that the target image block is an active image block, the noise state indicates that the target image block is a noisy image block, and the damaged state indicates that the target image block is not a damaged image block, then the evaluation value of the target image block is determined according to the noise quantization value corresponding to the target image block, and the quality evaluation result is determined according to each evaluation value.

[0105] In another possible implementation, if the active state indicates that the target image block is an active image block, the noise state indicates that the target image block is not a noise image block, and the damaged state indicates that the target image block is a damaged image block, then the evaluation value of the target image block is determined according to the damage quantization value corresponding to the target image block, and the quality evaluation result is determined according to each evaluation value.

[0106] The methods provided in this application are suitable for clinical and research applications, including quality assessment of dynamic PET, delayed PET, multi-frame PET, and pharmacokinetic PET images.

[0107] In the aforementioned medical image quality assessment method, the first statistical parameter, the second statistical parameter, and the third statistical parameter corresponding to the pixel values ​​of all pixels in the foreground image of the medical image to be assessed are obtained; for each target image block, the active state of the target image block is determined based on the first statistical parameter and the pixel values ​​corresponding to the target image block; the noise state of the target image block is determined based on the second statistical parameter and the pixel values ​​corresponding to the target image block; the damaged state of the target image block is determined based on the third statistical parameter and the pixel values ​​corresponding to the target image block; and the quality assessment result of the medical image to be assessed is determined based on the active state, noise state, and damaged state of each target image block. The first statistical parameter includes the mean, range, and preset percentile of all pixel values; the second statistical parameter includes the coefficient of variation; and the third statistical parameter includes the signal-to-noise ratio. The target image block is obtained by dividing the foreground image. This application's embodiments extract multi-dimensional statistical parameters from the foreground image, and determine the active state, noise state, and damaged state by combining the pixel values ​​of the target image block. Finally, the multi-state results are fused to determine the quality assessment result of the medical image to be evaluated. This not only specifically avoids the interference of the background area of ​​the medical image on the quality assessment, but also realizes the hierarchical assessment from global statistical features to local block-level states, so that the quality assessment results are consistent with the physical laws of medical imaging. At the same time, it takes into account the three quality indicators of active state, noise state, and damaged state, solves the problem that traditional algorithms cannot adapt to the characteristics of medical images, improves the objectivity, robustness, and clinical relevance of the quality assessment, and can adapt to the quality assessment needs of medical images under different scanning conditions and different imaging devices.

[0108] Figure 3 This is a flowchart illustrating an active state determination method in one embodiment, such as... Figure 3 As shown, this application embodiment relates to a possible implementation of how to determine the active state of a target image block based on a first statistical parameter and the pixel value corresponding to the target image block, including the following steps:

[0109] S301, determine the intensity and foreground factor of the target image block based on the pixel value corresponding to the target image block.

[0110] In this embodiment of the application, the average value of the pixel values ​​corresponding to the target image block is used as the intensity of the target image block; the area of ​​the foreground image in the target image block is obtained, and the foreground factor is determined according to the ratio of the area of ​​the foreground image in the target image block to the area of ​​the target image block.

[0111] S302, determine the candidate intensity threshold based on the relationship between the mean and the preset percentile, and compensate the candidate intensity threshold according to the range to obtain the intensity threshold.

[0112] The noise level of medical images varies significantly with the number of scans, scan duration, dosage, and reconstruction parameters. Algorithms with fixed thresholds are completely unsuitable for different scanning conditions. Therefore, in this embodiment, the intensity threshold is set to max(10th percentile, 0.1 × mean of all pixel values ​​in the foreground image) × range factor; the range factor is set to min(dynamic range / mean of all pixel values ​​in the foreground image, a preset constant). Optionally, the preset constant can be 5, 4, 7, etc.

[0113] S303, when the intensity is greater than the intensity threshold and the foreground factor is greater than the preset foreground factor threshold, the active state characterization target image block is determined as an active image block.

[0114] Optionally, the foreground factor thresholds are 0.9, 0.95, 0.85, 0.98, etc.

[0115] In this embodiment, the intensity and foreground factor of the target image block are determined based on the pixel values ​​corresponding to the target image block. A candidate intensity threshold is determined based on the relationship between the mean and a preset percentile. The candidate intensity threshold is then compensated based on the range to obtain the final intensity threshold. If the intensity is greater than the intensity threshold and the foreground factor is greater than a preset foreground factor threshold, the target image block representing the active state is determined to be an active image block. This embodiment combines the intensity and foreground factor of the target image block as dual-dimensional indicators, and dynamically determines the intensity threshold based on a first statistical parameter. This achieves accurate determination of the active state of the target image block, avoiding the one-sidedness of relying solely on pixel intensity to determine active regions. Furthermore, the foreground factor screening eliminates interference from non-living tissue regions, ensuring that the detected active image blocks are the true metabolically active structures in the image. This improves the accuracy of active state determination and lays a reliable positive indicator foundation for subsequent quality assessment by fusing multiple states, making the quality assessment more aligned with the core clinical diagnostic needs of medical images.

[0116] Figure 4 This is a flowchart illustrating a noise state determination method in one embodiment, such as... Figure 4 As shown, this application embodiment relates to a possible implementation of how to determine the noise state of a target image block based on a second statistical parameter and the pixel value corresponding to the target image block, including the following steps:

[0117] S401, Based on the pixel values ​​corresponding to the target image patch, determine the expected standard deviation and observed standard deviation of the target image patch.

[0118] In the embodiments of this application, the expected standard deviation It can be represented as , The mean and observation standard deviation of the target image patch. It can be represented as , where n is the number of pixels in the target image block.

[0119] S402, based on the expected standard deviation and the observed standard deviation, determine the excess noise ratio and Poisson perception bias of the target image patch.

[0120] Algorithms such as BRISQUE, NIQE, and PIQE are primarily designed for natural images, where noise is additive Gaussian noise. However, medical image noise exhibits a Poisson distribution and is strongly correlated with count intensity, often leading these methods to misclassify normal medical images as low-quality. Therefore, in this embodiment, the excess noise ratio (ENR) and Poisson perceptual bias of the target image patch are determined by combining the expected standard deviation and the observed standard deviation, demonstrating robustness to short scans, low doses, or high-noise images. The excess noise ratio (ENR) can be expressed as... And Poisson perception bias can be expressed as .

[0121] S403, determine the noise baseline of the target image patch based on the excess noise ratio, Poisson perception bias, and second statistical parameters.

[0122] In the embodiments of this application, ENR is first subjected to Normalization is performed by normalizing the coefficient of variation NL in the second statistical parameter to the range of 0-1, resulting in the normalized coefficient of variation NL_norm. NL_norm is used as the weighting coefficient for PAD, and 1-NL_norm is used as the weighting coefficient for the normalized ENR. The weighting coefficients are then used to perform weighted fusion of PAD and the normalized ENR to obtain the noise baseline of the target image patch.

[0123] In one possible implementation, a noise baseline for the target image patch is determined based on the excess noise ratio, Poisson perceptual bias, and a second statistical parameter, including:

[0124] If the coefficient of variation is greater than a preset threshold, a noise baseline is determined based on the first weighting coefficient, the excess noise ratio, and the Poisson perception bias; if the coefficient of variation is not greater than the preset threshold, a noise baseline is determined based on the second weighting coefficient, the third weighting coefficient, the excess noise ratio, and the Poisson perception bias.

[0125] Optionally, the preset threshold can be 0.5, 0.6, etc.

[0126] In this embodiment of the application, when the coefficient of variation is greater than a preset threshold, the following is adopted: A noise baseline is obtained, where 0.5 is the first weighting coefficient; if the coefficient of variation is not greater than a preset threshold, then... To determine the noise baseline, the second and third weighting coefficients can optionally be 0.6 and 0.4; 0.8 and 0.2; 0.9 and 0.1, etc. Based on the relationship between the coefficient of variation and the preset threshold, different weight combinations are used to fuse the excess noise ratio and Poisson perception bias, thereby determining the noise baseline. This allows the calculation of the noise baseline to dynamically adjust the parameter weights according to the global noise level (coefficient of variation), closely reflecting the differences in noise characteristics of medical images under different scanning conditions. In high global noise scenarios, it strengthens the representation of absolute noise deviation, and in low global noise scenarios, it strengthens the representation of relative noise proportion. This improves the adaptability of the noise baseline to medical images with different noise levels, making subsequent noise status determination based on the noise baseline more accurate and more in line with actual imaging conditions.

[0127] S404, determine the noise state of the target image patch based on the observation standard deviation and noise baseline.

[0128] In this embodiment, the observation standard deviation and the noise benchmark can be directly compared. If the observation standard deviation is greater than the noise benchmark, the target image patch representing the noise state is determined to be a noise image patch.

[0129] In one possible implementation, the noise state of the target image block is determined based on the observation standard deviation and the noise benchmark, including: determining the product of the noise benchmark and the preset noise value; and, if the observation standard deviation is greater than the product, determining the noise state of the target image block to characterize the target image block as a noisy image block.

[0130] Optionally, the preset noise value is a random value between 0 and 2.

[0131] In the embodiments of this application, if Then, the noise state of the target image patch is determined to be a noisy image patch, where 2×S noise The preset noise value is S. noise The value is a random value between 0 and 1. In this embodiment, the noise threshold is determined by multiplying a noise benchmark by a preset noise value. The observed standard deviation is compared with this threshold to determine the noise state, thus constructing a clear and quantifiable noise determination standard. By dynamically changing the noise threshold through the preset noise value, the adaptability of the noise threshold under different global noise levels is ensured, allowing the algorithm to automatically adapt to different imaging devices and scanning parameters. For example, it can adapt to different scanning times, doses, and count levels.

[0132] In this embodiment, the expected standard deviation and observed standard deviation of the target image block are determined based on the pixel values ​​corresponding to the target image block. Based on these, the excess noise ratio and Poisson perceptual bias of the target image block are determined. Based on the excess noise ratio, Poisson perceptual bias, and a second statistical parameter, the noise baseline of the target image block is determined. Finally, based on the observed standard deviation and the noise baseline, the noise state of the target image block is determined. This embodiment, based on the inherent characteristics of Poisson noise in medical images, calculates the expected and observed standard deviations of the target image block, further derives the excess noise ratio and Poisson perceptual bias, and then combines the second statistical parameter to determine the noise baseline and noise state. This achieves accurate detection of noise in medical images, overcoming the limitations of traditional algorithms that assume noise is Gaussian noise. It effectively distinguishes between normal statistical noise and abnormally high noise in medical images, while linking the noise state determination with the global noise level, avoiding misjudgments from single-block local analysis, improving the adaptability and accuracy of noise state determination, and providing a reliable noise dimension index for medical image quality assessment.

[0133] Figure 5 This is a flowchart illustrating a method for determining the damaged state in one embodiment, such as... Figure 5 As shown, this application embodiment relates to a possible implementation of how to determine the damage state of a target image patch based on a third statistical parameter and the pixel value corresponding to the target image patch, including the following steps:

[0134] S501, divide the target image block into multiple line segments by dividing each border.

[0135] In this embodiment of the application, the borders of the target image block are divided based on a preset number of line segments to obtain multiple line segments. Each line segment may or may not contain duplicate pixels.

[0136] When dividing the borders based on a preset number, the borders can be divided evenly or unevenly. For example, if the preset number is 16 and the target image block size is 16×16, taking excluding duplicate pixels as an example, each border can be divided into 4 segments, each segment containing 4 pixels, for a total of 16 segments; alternatively, the first segment could contain 3 pixels, the second segment 4 pixels, the third segment 5 pixels, and so on.

[0137] In one possible implementation, the number of line segments can also be obtained by subtracting the size of the preset window from the size of each border of the target image block and adding 1; for example, the number of line segments = 16 - 6 + 1 = 11 segments, and each line segment includes 6 pixels.

[0138] S502 determines the segment standard deviation corresponding to each line segment based on the pixel values ​​on the line segment.

[0139] In the embodiments of this application, for each line segment, the segmental standard deviation corresponding to each line segment can be calculated using the above-described method for calculating the observation standard deviation.

[0140] S503, determine the threshold for damaged blocks based on the third statistical parameter.

[0141] In the embodiments of this application, the damaged block threshold can be calculated in the following way: damaged block threshold = 0.05 × (1 + 1 / (signal-to-noise ratio + 0.1)).

[0142] S504, if at least one segment standard deviation is greater than or equal to the damaged block threshold, the target image block representing the damaged state is determined to be a damaged image block.

[0143] A damaged block is defined as a segment on the edge whose standard deviation is significantly smaller than the standard deviation of the entire image block. However, this is clearly incorrect for medical images, as local activity is relatively uniform, and there will inevitably be some regions with a standard deviation smaller than the standard deviation of the entire image block. Therefore, this application defines a target image block as a damaged image block if at least one segment's standard deviation is greater than or equal to a damaged block threshold.

[0144] In this embodiment, the borders of the target image block are divided into multiple line segments. For each line segment, the segmental standard deviation is determined based on the pixel values ​​on the line segment. A damaged block threshold is determined based on a third statistical parameter. If at least one segmental standard deviation is greater than or equal to the damaged block threshold, the target image block is determined to be a damaged image block. This embodiment calculates the segmental standard deviation of each border segment of the target image block and determines the damaged block threshold using the third statistical parameter. The determination of the damaged state is based on whether the segmental standard deviation exceeds the threshold, achieving accurate detection of structural damage in medical image blocks. This effectively solves the problem that traditional algorithms cannot detect metabolic structural distortion in medical images, improves the accuracy of damaged state determination, and links the determination result with the quality of the global image, enhancing the robustness of the algorithm.

[0145] Figure 6 This is a flowchart illustrating a method for determining quality assessment results in one embodiment, such as... Figure 6 As shown, this application embodiment relates to a possible implementation method for determining the quality assessment result of a medical image to be evaluated based on the activity state, noise state, and damage state of each target image patch, including the following steps:

[0146] S601, when the target image block is characterized as a noisy image block, the variance of the target image block is determined as the noise quantization value of the target image block.

[0147] S602, when the target image block is characterized as a damaged image block, the maximum variance corresponding to each line segment of the target image block is determined as the damage quantization value of the target image block.

[0148] In this embodiment of the application, the variance of all pixels on each line segment is obtained, and the maximum variance is used as the damage quantization value of the target image block.

[0149] S603, when the target image block is characterized as an active image block in the active state, the evaluation value of the target image block is determined based on the noise quantization value and / or the damaged quantization value.

[0150] High-activity structures have low artifacts, reasonable noise (neither too high nor too low), and normal tissue edge structures. Therefore, in this application, when the target image block is characterized as an active image block in an active state, the evaluation value of the target image block is determined based on the noise quantization value and the damage quantization value.

[0151] In this embodiment, the evaluation value of the target image block can be expressed as: blockScore = WHSA×WNDC×(1−max( , ))+ WHSA×WNC×max( , WHSA indicates the active state; if the target image patch is active, WHSA equals 1. WNDC indicates the damaged state; if the target image patch is damaged, WNDC equals 1. WNC indicates the noisy state; if the target image patch is noisy, WNC equals 1. This is the noise quantization value. This is the quantified value of the damage.

[0152] S604, determine the quality assessment result based on the evaluation value of each target image patch.

[0153] In this embodiment of the application, a target image block whose evaluation value is less than a preset evaluation value can be identified as an image block to be evaluated, and the quality evaluation result can be determined based on the evaluation value of the image block to be evaluated and the number of image blocks to be evaluated.

[0154] In one possible implementation, the quality assessment result can also be determined directly based on the evaluation value of each target image patch and the number of target image patches. For example, the average value of the evaluation values ​​can be determined based on the evaluation value of each target image patch and the number of target image patches, and the average value of the evaluation values ​​can be used as the quality assessment result.

[0155] In this embodiment, when the target image block is characterized as a noisy image block, the variance of the target image block is determined as the noise quantization value. When the target image block is characterized as a damaged image block, the maximum variance corresponding to each line segment of the target image block is determined as the damage quantization value. When the target image block is characterized as an active image block, the quality assessment result is determined based on the noise quantization value and the damage quantization value. This embodiment determines corresponding quantization indicators for target image blocks in different states, and fuses the noise quantization value and the damage quantization value for active image blocks to determine the quality assessment result. This achieves the fusion of multi-dimensional features and eliminates the interference of inactive blocks on the assessment result. The quality assessment result better reflects the true and effective quality of medical images and improves the clinical reference value of the assessment result.

[0156] Figure 7 This is a flowchart illustrating a method for determining quality assessment results in another embodiment, as shown below. Figure 7 As shown, this application embodiment relates to a possible implementation of how to determine the quality assessment result of a medical image to be evaluated based on the evaluation values ​​of each target image patch, including the following steps:

[0157] S701, the target image block whose evaluation value is less than the preset evaluation value is determined as the image block to be evaluated.

[0158] Optionally, the preset evaluation value can be 100, 90, etc.

[0159] S702, determine the quality assessment result based on the evaluation value of the image patch to be evaluated and the number of image patches to be evaluated.

[0160] In this embodiment, the evaluation values ​​of the image blocks to be evaluated are summed, and the ratio of the summation to the number of image blocks to be evaluated is used as the quality evaluation result. By filtering out image blocks whose evaluation values ​​are less than a preset value, the quality evaluation result is finally determined by combining the evaluation values ​​and the number of image blocks to be evaluated. This avoids the influence of individual abnormal blocks on the overall quality evaluation result, improves the stability and reliability of the evaluation result, and adapts to the needs of automated quality evaluation of large batches of medical images.

[0161] Further, based on the evaluation values ​​of the image blocks to be evaluated and the number of image blocks to be evaluated, the quality evaluation result is determined, including: determining the sum of the evaluation values ​​of each image block to be evaluated, and a first summation value with a preset constant value; determining the number of image blocks to be evaluated and a second summation value with a preset constant value; and determining the quality evaluation result based on the ratio of the first summation value and the second summation value.

[0162] In this embodiment of the application, the quality assessment result can be expressed as: Where NHSA is the number of image patches to be evaluated and C is a preset constant value.

[0163] In this embodiment, target image blocks with evaluation values ​​less than a preset evaluation value are identified as image blocks to be evaluated. A first summation value is determined by combining the sum of the evaluation values ​​of each image block to be evaluated with a preset constant value. A second summation value is determined by combining the number of image blocks to be evaluated with a preset constant value. The quality evaluation result is determined based on the ratio of the first and second summation values. This embodiment improves the stability and robustness of the quality evaluation result by calculating the first summation value of the evaluation values ​​of the image blocks to be evaluated with a preset constant, and the second summation value of the number of image blocks to be evaluated with a preset constant value, and finally determining the quality evaluation result based on their ratio.

[0164] In one exemplary embodiment, such as Figures 8-23 As shown, Figures 8-23 By combining PET scan experiments with different scan durations, equipment parameters, and tracer metabolism times, the effectiveness of this quality assessment method was visually verified. Medical images presented the actual quality performance of PET slices under different experimental conditions, and bar charts quantified the corresponding quality assessment results. The lower the image quality assessment result, the better the image quality. Figures 8-11 The image quality assessment results are shown for different scan durations of 10 min, 20 min, and 30 min. Figures 12-15 The image quality assessment results are shown for different scan durations of 6s, 10min, and 20min. Figures 16-19 The image quality assessment results under different device parameters are shown. Figures 20-23 The results of image quality assessment under different tracer metabolism times are presented, demonstrating that the method can accurately distinguish PET images of different quality levels, and the scoring results are highly consistent with the actual quality patterns of clinical visual assessment and imaging conditions.

[0165] It should be understood that although the steps in the flowcharts of the embodiments described above are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the embodiments described above may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages in other steps. It is understood that the steps in different embodiments can be freely combined as needed, and all non-contradictory solutions formed by such combinations are within the scope of protection of this application.

[0166] Based on the same inventive concept, this application also provides a medical image quality assessment device for implementing the medical image quality assessment method described above. The solution provided by this device is similar to the solution described in the above method; therefore, the specific limitations in one or more embodiments of the medical image quality assessment device provided below can be found in the limitations of the medical image quality assessment method described above, and will not be repeated here.

[0167] In one exemplary embodiment, such as Figure 24 As shown, a medical image quality assessment device is provided, comprising: an acquisition module 11, a first determination module 12, a second determination module 13, a third determination module 14, and a fourth determination module 15, wherein:

[0168] The acquisition module 11 is used to acquire the first statistical parameter, the second statistical parameter and the third statistical parameter corresponding to the pixel values ​​of all pixels in the foreground image of the medical image to be evaluated; the first statistical parameter includes the mean, range and preset percentile of all pixel values, the second statistical parameter includes the coefficient of variation and the third statistical parameter includes the signal-to-noise ratio.

[0169] The first determining module 12 is used to determine the active state of each target image block based on the first statistical parameters and the pixel values ​​corresponding to the target image block; the target image block is obtained by dividing the foreground image.

[0170] The second determining module 13 is used to determine the noise state of the target image block based on the second statistical parameters and the pixel values ​​corresponding to the target image block;

[0171] The third determining module 14 is used to determine the damage state of the target image block based on the third statistical parameter and the pixel value corresponding to the target image block;

[0172] The fourth determination module 15 is used to determine the quality assessment result of the medical image to be evaluated based on the active state, noise state, and damaged state of each target image block.

[0173] In an exemplary embodiment, the first determining module 12 is specifically used to determine a candidate intensity threshold based on the relationship between the mean and the preset percentile, and to compensate the candidate intensity threshold based on the range to obtain an intensity threshold; when the intensity is greater than the intensity threshold and the foreground factor is greater than the preset foreground factor threshold, the active state characterizing target image block is determined to be an active image block.

[0174] In an exemplary embodiment, the second determining module 13 is specifically configured to: determine the expected standard deviation and observed standard deviation of the target image block based on the pixel values ​​corresponding to the target image block; determine the excess noise ratio and Poisson perceptual bias of the target image block based on the expected standard deviation and observed standard deviation; determine the noise baseline of the target image block based on the excess noise ratio, Poisson perceptual bias and second statistical parameter; and determine the noise state of the target image block based on the observed standard deviation and the noise baseline.

[0175] In an exemplary embodiment, the second determining module 13 is specifically used to determine the product of the noise benchmark and the preset noise value; if the observed standard deviation is greater than the product result, the noise state characterization of the target image block is determined to be a noisy image block.

[0176] In an exemplary embodiment, the second determining module 13 is specifically used to determine a noise benchmark based on a first weighting coefficient, an excess noise ratio, and a Poisson perception bias when the coefficient of variation is greater than a preset threshold; and to determine a noise benchmark based on a second weighting coefficient, a third weighting coefficient, an excess noise ratio, and a Poisson perception bias when the coefficient of variation is not greater than the preset threshold.

[0177] In an exemplary embodiment, the third determining module 14 is specifically used to divide each border of the target image block into multiple line segments; for each line segment, determine the segment standard deviation corresponding to the line segment based on the pixel value on the line segment; determine the damaged block threshold based on the third statistical parameter; and determine the damaged state characterizing target image block as a damaged image block if at least one segment standard deviation is greater than or equal to the damaged block threshold.

[0178] In an exemplary embodiment, the fourth determining module 15 is specifically configured to: determine the variance of the pixel values ​​of all pixels in the target image block as the noise quantization value of the target image block when the target image block is characterized as a noisy image block in a noise state; determine the maximum variance of the pixel values ​​of all pixels on each line segment of the target image block as the damage quantization value of the target image block when the target image block is characterized as a damaged image block in a damaged state; determine the evaluation value of the target image block based on the noise quantization value and / or the damage quantization value when the target image block is characterized as an active image block in an active state; and determine the quality evaluation result based on the evaluation value of each target image block.

[0179] In an exemplary embodiment, the fourth determining module 15 is specifically used to determine the target image block whose evaluation value is less than the preset evaluation value as the image block to be evaluated; and to determine the quality evaluation result based on the evaluation value of the image block to be evaluated and the number of image blocks to be evaluated.

[0180] In an exemplary embodiment, the fourth determining module 15 is specifically used to determine the sum of the evaluation values ​​of each image block to be evaluated, and a first sum of the values ​​with a preset constant value; to determine the number of image blocks to be evaluated and a second sum of the values ​​with a preset constant value; and to determine the quality evaluation result based on the ratio of the first sum of the values ​​and the second sum of the values.

[0181] Each module in the aforementioned medical image quality assessment device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device, or stored in the memory of a computer device as software, so that the processor can call and execute the corresponding operations of each module.

[0182] In one exemplary embodiment, a computer device is provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps of any of the above method embodiments.

[0183] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a processor, implements the steps of any of the above method embodiments.

[0184] In one embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements the steps of any of the above method embodiments.

[0185] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of the relevant data must comply with relevant regulations.

[0186] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile memory and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, artificial intelligence (AI) processors, etc., and are not limited to these.

[0187] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this application.

[0188] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. A method for assessing the quality of medical images, characterized in that, The method includes: Obtain a first statistical parameter, a second statistical parameter, and a third statistical parameter corresponding to the pixel values ​​of all pixels in the foreground image of the medical image to be evaluated; the first statistical parameter includes the mean, range, and preset percentile of all pixel values, the second statistical parameter includes the coefficient of variation, and the third statistical parameter includes the signal-to-noise ratio; For each target image block, the active state of the target image block is determined based on the first statistical parameter and the pixel value corresponding to the target image block; the target image block is obtained by dividing the foreground image. The noise state of the target image block is determined based on the second statistical parameter and the pixel value corresponding to the target image block; The damage state of the target image block is determined based on the third statistical parameter and the pixel value corresponding to the target image block; The quality assessment result of the medical image to be evaluated is determined based on the active state, noise state, and damaged state of each target image block.

2. The method according to claim 1, characterized in that, Determining the activity state of the target image block based on the first statistical parameter and the pixel value corresponding to the target image block includes: The intensity and foreground factor of the target image block are determined based on the pixel values ​​corresponding to the target image block. Based on the relationship between the mean and the preset percentile, a candidate intensity threshold is determined, and the candidate intensity threshold is compensated based on the range to obtain the intensity threshold. When the intensity is greater than the intensity threshold and the foreground factor is greater than the preset foreground factor threshold, the active state is determined to characterize the target image block as an active image block.

3. The method according to claim 1, characterized in that, Determining the noise state of the target image block based on the second statistical parameter and the pixel value corresponding to the target image block includes: Based on the pixel values ​​corresponding to the target image patch, determine the expected standard deviation and observed standard deviation of the target image patch; The excess noise ratio and Poisson perception bias of the target image patch are determined based on the expected standard deviation and the observed standard deviation. The noise baseline of the target image patch is determined based on the excess noise ratio, the Poisson perception bias, and the second statistical parameter. The noise state of the target image patch is determined based on the observation standard deviation and the noise benchmark.

4. The method according to claim 3, characterized in that, Determining the noise state of the target image patch based on the observation standard deviation and the noise benchmark includes: Determine the product of the noise reference and the preset noise value; If the observed standard deviation is greater than the product result, the noise state of the target image block is determined to characterize the target image block as a noisy image block.

5. The method according to claim 3, characterized in that, Determining the noise baseline of the target image patch based on the excess noise ratio, the Poisson perceptual bias, and the second statistical parameter includes: If the coefficient of variation is greater than a preset threshold, the noise benchmark is determined based on the first weighting coefficient, the excess noise ratio, and the Poisson perception bias. If the coefficient of variation is not greater than the preset threshold, the noise benchmark is determined based on the second weighting coefficient, the third weighting coefficient, the excess noise ratio, and the Poisson perception bias.

6. The method according to claim 1, characterized in that, Determining the damage state of the target image patch based on the third statistical parameter and the pixel value corresponding to the target image patch includes: The target image block is divided into multiple line segments by dividing each border. For each line segment, the segment standard deviation corresponding to the line segment is determined based on the pixel values ​​on the line segment; The threshold for damaged blocks is determined based on the third statistical parameter. If at least one of the segment standard deviations is greater than or equal to the damaged block threshold, the damaged state characterizes the target image block as a damaged image block.

7. The method according to claim 1, characterized in that, The step of determining the quality assessment result of the medical image to be evaluated based on the activity state, noise state, and damage state of each target image patch includes: When the noise state characterizes the target image block as a noisy image block, the variance corresponding to the pixel values ​​of all pixels in the target image block is determined as the noise quantization value of the target image block; When the damaged state characterizes the target image block as a damaged image block, the maximum variance corresponding to the pixel values ​​of all pixels on each line segment of the target image block is determined as the damage quantization value of the target image block; When the active state indicates that the target image patch is an active image patch, the evaluation value of the target image patch is determined based on the noise quantization value and / or the damaged quantization value. The quality assessment result is determined based on the evaluation value of each target image patch.

8. The method according to claim 7, characterized in that, Determining the quality assessment result based on the evaluation value of each target image patch includes: The target image block whose evaluation value is less than the preset evaluation value is identified as the image block to be evaluated; The quality assessment result is determined based on the evaluation value of the image patch to be evaluated and the number of image patches to be evaluated.

9. A medical image quality assessment device, characterized in that, The device includes: The acquisition module is used to acquire a first statistical parameter, a second statistical parameter, and a third statistical parameter corresponding to the pixel values ​​of all pixels in the foreground image of the medical image to be evaluated; the first statistical parameter includes the mean, range, and preset percentile of all pixel values, the second statistical parameter includes the coefficient of variation, and the third statistical parameter includes the signal-to-noise ratio; The first determining module is used to determine the active state of each target image block based on the first statistical parameters and the pixel values ​​corresponding to the target image block; the target image block is obtained by dividing the foreground image. The second determining module is used to determine the noise state of the target image block based on the second statistical parameters and the pixel values ​​corresponding to the target image block; The third determining module is used to determine the damage state of the target image block based on the third statistical parameter and the pixel value corresponding to the target image block; The fourth determining module is used to determine the quality assessment result of the medical image to be evaluated based on the active state, noise state, and damaged state of each target image block.

10. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1-8.

Images