Image noise reduction method, device, computer equipment and storage medium

Abstract

The present application relates to an image denoising method, device, computer equipment and storage medium, which obtains a low signal-to-noise ratio image and a high signal-to-noise ratio image of a detection object, obtains a difference image between the low signal-to-noise ratio image and the high signal-to-noise ratio image, extracts an additional information image including non-noise information from the difference image, and corrects the high signal-to-noise ratio image of the detection object according to the non-noise information in the additional information image to obtain a corrected high signal-to-noise ratio image. The method can improve the quality of the denoised image.

Description

Technical Field

[0001] The present application relates to the field of image processing technology, and in particular to an image noise reduction method, device, computer equipment and storage medium. Background Art

[0002] In the field of image applications, image quality is of vital importance. For example, during magnetic resonance scanning, due to the influence of the imaging mechanism (including data acquisition, image reconstruction method, etc.), noise will be introduced into the final image, thereby reducing the image quality, blurring tissue boundaries and making fine structures difficult to identify, thus affecting diagnosis.

[0003] Existing methods for removing noise from magnetic resonance images mainly use various algorithms to reduce the noise level of the image while retaining detail information as much as possible, such as image domain denoising and transform domain denoising; or, denoising through a sufficiently trained deep learning network.

[0004] However, the current algorithm has limited noise reduction capabilities and often introduces other additional information while reducing noise, resulting in loss of image details or image errors after noise reduction, making the quality of the image after noise reduction poor. Summary of the invention

[0005] Based on this, it is necessary to provide an image denoising method, device, computer equipment and storage medium to address the above technical problems, which can ensure that the denoised image does not lose details or make errors, and improve the quality of the denoised image.

[0006] In a first aspect, an embodiment of the present application provides an image noise reduction method, the method comprising:

[0007] Acquire low signal-to-noise ratio images and high signal-to-noise ratio images of the detection object;

[0008] Obtaining a difference image between a low signal-to-noise ratio image and a high signal-to-noise ratio image of the detection object;

[0009] Extracting an additional information image from the difference image; the additional information image includes non-noise information in the difference image;

[0010] According to the non-noise information, the high signal-to-noise ratio image of the detection object is corrected to obtain a corrected high signal-to-noise ratio image.

[0011] In one embodiment, obtaining a low signal-to-noise ratio image and a high signal-to-noise ratio image of a detection object includes:

[0012] Through the preset deep learning denoising network, the low signal-to-noise ratio image of the detection object is denoised to obtain a high signal-to-noise ratio image of the detection object.

[0013] In one embodiment, obtaining a difference image between a low signal-to-noise ratio image and a high signal-to-noise ratio image of a detection object includes:

[0014] Subtract the image with low signal-to-noise ratio from the image with high signal-to-noise ratio to obtain the difference image.

[0015] In one of the embodiments, before extracting the additional information image from the difference image, the method further comprises:

[0016] Detect whether there is non-noise information in the difference image;

[0017] If non-noise information exists, a step of extracting an additional information image from the difference image is performed.

[0018] In one embodiment, detecting whether there is non-noise information in the difference image includes:

[0019] Obtaining signal distribution in the difference image;

[0020] If the signal distribution all appears as noise, it is determined that there is no non-noise information in the difference image;

[0021] If the signal distribution contains information other than noise, it is determined that non-noise information exists in the difference image.

[0022] In one embodiment, obtaining a difference image between a low signal-to-noise ratio image and a high signal-to-noise ratio image of the detected object further includes:

[0023] Divide the detection object into multiple sub-areas;

[0024] Obtain a difference image between the low signal-to-noise ratio image and the high signal-to-noise ratio image of each sub-region.

[0025] In one embodiment, the method further comprises:

[0026] The non-noise information in the difference image of each sub-region is extracted, and the high signal-to-noise ratio image of each sub-region is corrected according to the extracted non-noise information to obtain the corrected high signal-to-noise ratio image of each sub-region.

[0027] In a second aspect, an embodiment of the present application provides an image noise reduction device, the device comprising:

[0028] The acquisition module is used to acquire a low signal-to-noise ratio image and a high signal-to-noise ratio image of the detection object.

[0029] A difference-finding module is used to obtain a difference image between a low signal-to-noise ratio image and a high signal-to-noise ratio image of the detection object;

[0030] An extraction module is used to extract an additional information image from the difference image; the additional information image includes non-noise information in the difference image;

[0031] The correction module is used to correct the high signal-to-noise ratio image of the detection object according to the non-noise information to obtain a corrected high signal-to-noise ratio image.

[0032] In a third aspect, an embodiment of the present application provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the steps of any method provided in the embodiment of the first aspect are implemented.

[0033] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the steps of any method provided in the embodiment of the first aspect above are implemented.

[0034] The embodiment of the present application provides an image denoising method, device, computer equipment and storage medium, which obtains a low signal-to-noise ratio image and a high signal-to-noise ratio image of a detection object, obtains a difference image between the low signal-to-noise ratio image and the high signal-to-noise ratio image, extracts an additional information image including non-noise information from the difference image, and corrects the high signal-to-noise ratio image of the detection object according to the non-noise information in the additional information image to obtain a corrected high signal-to-noise ratio image. Since the additional information image includes non-noise information between the high signal-to-noise ratio image and the low signal-to-noise ratio image, correcting the high signal-to-noise ratio image of the detection object according to the non-noise information is equivalent to removing the non-noise information in the high signal-to-noise ratio image, and the remaining information is correct. In this way, the corrected high signal-to-noise ratio image is the high signal-to-noise ratio image after removing the non-noise information, thereby improving the quality of the image after denoising. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] Figure 1 A diagram showing an application environment of an image noise reduction method in an embodiment;

[0036] Figure 2 is a schematic diagram of a flow chart of an image noise reduction method in one embodiment;

[0037] Figure 3 A schematic diagram of an excitation sequence in one embodiment;

[0038] Figure 4 A schematic diagram of an excitation sequence in another embodiment;

[0039] Figure 5 is a schematic flow chart of an image noise reduction method in another embodiment;

[0040] Figure 6 is a schematic flow chart of an image noise reduction method in another embodiment;

[0041] Figure 7 is a schematic flow chart of an image noise reduction method in another embodiment;

[0042] Figure 8 is a flow chart of an image noise reduction method in one embodiment;

[0043] Fig. 9 is a schematic diagram of an image denoising method in one embodiment;

[0044] Fig.10 FIG. 4 is a structural frame of an image noise reduction device in an embodiment. DETAILED DESCRIPTION

[0045] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0046] The image denoising method provided in this application can be applied to Figure 1 In the application environment shown in FIG. , the image denoising method is applied in Figure 1 In the computer device shown, the computer device may be a server, and its internal structure diagram can be seen in Figure 1 . The computer device includes a processor, a memory and a network interface connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store relevant data of image noise reduction. The network interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a method for image noise reduction is implemented.

[0047] The embodiments of the present application provide an image denoising method, apparatus, computer equipment and storage medium, which can improve the quality of the denoised image. The technical solution of the present application and how the technical solution of the present application solves the above-mentioned technical problems will be described in detail below through embodiments and in combination with the accompanying drawings. The following specific embodiments can be combined with each other, and the same or similar concepts or processes may not be repeated in some embodiments. It should be noted that the image denoising method provided by the present application, Figure 2-Figure 9 The execution subject is a computer device. Figure 2-Figure 9 The execution subject may also be an image noise reduction device, which may be implemented as part or all of a computer device through software, hardware, or a combination of software and hardware.

[0048] In order to make the purpose, technical solution and advantages of the embodiments of the present application clearer, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all of the embodiments. In one embodiment, Figure 2 As shown, an image denoising method is provided, which is applied to Figure 1 The computer device in the embodiment is used as an example to illustrate that the present embodiment relates to a specific process in which the computer device corrects the high signal-to-noise ratio image of the detection object after extracting the additional information image based on the difference image between the low signal-to-noise ratio image and the high signal-to-noise ratio image. The embodiment includes the following steps:

[0049] S101, obtaining a low signal-to-noise ratio image and a high signal-to-noise ratio image of a detection object, and obtaining a difference image between the low signal-to-noise ratio image and the high signal-to-noise ratio image of the detection object.

[0050] Optionally, the pixel value of each pixel point in the difference image is obtained by subtracting the pixel value of each pixel point in the low signal-to-noise ratio image from the pixel value of the pixel point in the high signal-to-noise ratio image.

[0051] The detection object may be a living organism such as an animal or human body, and may be an organ or tissue local area such as the heart, lung, artery, liver, pancreas, kidney, uterus, breast, etc. of the organism, and the low signal-to-noise ratio image and / or the high signal-to-noise ratio image may be obtained by performing a magnetic resonance scan, a CT (computed tomography) scan, a PET (positron emission tomography)-MR (magnetic resonance) scan, etc. on the detection object, and this embodiment does not limit this. The low signal-to-noise ratio image indicates that the effective signal of the detection object image in the image is relatively low (compared to the high signal-to-noise ratio image) and the random noise is high; the high signal-to-noise ratio image indicates that the effective signal of the detection object in the image is relatively high and the random noise is low.

[0052] In one embodiment, the low signal-to-noise ratio image and the high signal-to-noise ratio image of the detection object are obtained through an image post-processing process: after the computer device obtains the low signal-to-noise ratio image of the detection object, it obtains the high signal-to-noise ratio image of the detection object through a noise reduction algorithm, and then obtains the difference image between the high signal-to-noise ratio image and the low signal-to-noise ratio image. In practical applications, the computer device obtains the low signal-to-noise ratio image of the detection object by directly using the original image of the detection object as the low signal-to-noise ratio image, or by receiving the low signal-to-noise ratio image of the detection object sent by other devices; wherein the noise reduction algorithm can be performed by the computer device itself or by other devices, including but not limited to Gaussian filtering, median filtering, wavelet threshold denoising, PM equation denoising in the image domain or frequency domain, and can also be one of the methods such as using a trained neural network, or a combination of two or more noise reduction algorithms in the aforementioned methods, so as to achieve a better noise reduction effect.

[0053] S102, extracting an additional information image from the difference image; the additional information image includes non-noise information in the difference image.

[0054] After obtaining the difference image, the computer device extracts an additional information image from the difference image, where the additional information refers to the invalid denoising information or non-noise information of the high signal-to-noise ratio image relative to the low signal-to-noise ratio image.

[0055] Taking the example of using a denoising method to process a low signal-to-noise ratio image to obtain a high signal-to-noise ratio image, the additional information is the useless and erroneous information introduced between the low signal-to-noise ratio image and the high signal-to-noise ratio image of the detection object, or the additional erroneous information introduced by the denoising process or the loss of image details; it should be noted that the additional information does not include noise, that is, it includes erroneous information other than the noise removal. For example, after denoising image A to obtain image B, there is some additional non-noise information in the denoised image B. This information has no effect on the content of the image and will affect the judgment of the image content, so this erroneous information should be removed.

[0056] For example, the computer device may extract the additional information image from the difference image by using a pre-trained neural network model, that is, inputting the difference image into the additional information extraction network model to obtain the additional information image. Alternatively, the computer device may shield all noise information, image content-related information, etc. in the difference image, and determine the remaining information as additional information to obtain the additional information image; or the computer device may extract the additional non-noise information by wavelet transform, contourlet (non-adaptive directional multi-scale analysis method) transform, etc. This embodiment does not limit the extraction method.

[0057] S103, correcting the high signal-to-noise ratio image of the detection object according to the non-noise information to obtain a corrected high signal-to-noise ratio image.

[0058] After obtaining the additional information image in the difference image, that is, obtaining the non-noise information between the high signal-to-noise ratio image and the low signal-to-noise ratio image, the computer device corrects the high signal-to-noise ratio image of the detection object according to the additional information image to obtain a corrected high signal-to-noise ratio image. For example, the erroneous information or non-noise information in the high signal-to-noise ratio image is removed, so that the remaining information is correct, thereby obtaining a high signal-to-noise ratio image with the erroneous information removed.

[0059] The image denoising method provided in this embodiment obtains a low signal-to-noise ratio image and a high signal-to-noise ratio image of the detection object, obtains a difference image between the low signal-to-noise ratio image and the high signal-to-noise ratio image, extracts an additional information image including non-noise information from the difference image, and corrects the high signal-to-noise ratio image of the detection object according to the non-noise information in the additional information image to obtain a corrected high signal-to-noise ratio image. Since the additional information image includes non-noise information between the high signal-to-noise ratio image and the low signal-to-noise ratio image, correcting the high signal-to-noise ratio image of the detection object according to the non-noise information is equivalent to removing the non-noise information in the high signal-to-noise ratio image, and the remaining information is correct. In this way, the corrected high signal-to-noise ratio image is the high signal-to-noise ratio image after removing the non-noise information, thereby improving the quality of the denoised image. Further, when applied to the medical field, the medical image corrected by this method can ensure that the denoised image does not lose details or increase erroneous information compared to the original low signal-to-noise ratio image, thereby improving the accuracy of diagnosis.

[0060] Optionally, when acquiring a low signal-to-noise ratio image and a high signal-to-noise ratio image of the detection object, after acquiring the low signal-to-noise ratio image, noise reduction is performed on the low signal-to-noise ratio image of the detection object to obtain a high signal-to-noise ratio image of the detection object.

[0061] Assume that the low signal-to-noise ratio image of the detection object is the original image of the detection object. After the computer device obtains the original image of the detection object, the computer device performs noise reduction on the original image to obtain a high signal-to-noise ratio image of the detection object. The noise reduction method includes but is not limited to various noise reduction algorithms, noise reduction networks trained by deep learning, etc., which are not limited to this. The signal-to-noise ratio refers to the ratio of the effective signal of the image to the random noise. The higher the signal-to-noise ratio, the higher the effective signal of the image and the lower the random noise.

[0062] Optionally, taking a denoising network trained by deep learning as an example, denoising is performed on a low signal-to-noise ratio image of the detection object to obtain a high signal-to-noise ratio image of the detection object, including: denoising the low signal-to-noise ratio image of the detection object through a preset deep learning denoising network to obtain a high signal-to-noise ratio image of the detection object.

[0063] A denoising network is trained in advance using deep learning technology, wherein the denoising network can be trained using a large number of diverse training samples, so that the trained denoising network can be more robust and accurate. When applied, the computer device inputs a low signal-to-noise ratio image of the detection object into the denoising network, and a high signal-to-noise ratio image of the detection object can be directly obtained.

[0064] Among them, the training method of the deep learning neural network uses the image obtained by scanning the scanned object at a low field strength as the input data of the deep learning neural network, and uses the image obtained by scanning the scanned object at a high field strength as the label data of the deep learning neural network. Exemplarily, the image obtained by a single excitation scan of a 0.5T MR device is used as the input data, and the image obtained by a single excitation scan of a 1.5T MR device is used as the label data, and its signal-to-noise ratio is equivalent to the signal-to-noise ratio of the image obtained by nine excitation scans of a 0.5T MR device. According to the input data and the label data, the parameters of each node of the deep learning neural network are adjusted.

[0065] The training method of the deep learning neural network uses the images of the scanned object obtained under different scanning sequences as training data. In one embodiment, the images of the scanned object obtained by the Balance-SSFP sequence are the input data of the deep learning neural network, and the images of the scanned object obtained by the dual-excitation Balance-SSFP sequence are the label data of the deep learning neural network.

[0066] like Figure 3 As shown, it is a schematic diagram of the Balance-SSFP sequence used for the first excitation in the embodiment of the present application, wherein: RF represents the radio frequency excitation pulse; Gss represents the layer selection gradient; Gpe represents the phase encoding gradient; Gro represents the readout frequency encoding gradient. TR represents the time interval between two adjacent RFs; TE represents the time interval from the midpoint of the RF that generates the macroscopic transverse magnetization vector to the midpoint of the collected echo signal. In this embodiment, a positive phase RF pulse α+ and a negative phase RF pulse α- are applied between TRs. The application time of Gss is the same as that of α+, and then Gpe and Gro are applied respectively. After TE, a gradient field with the same magnitude and opposite direction as the corresponding spatial encoding gradient field is applied in the layer selection direction, phase encoding direction, and readout frequency encoding direction. The magnetic resonance signal collected during the TE period corresponds to the input data.

[0067] like Figure 4As shown, it is the Balance-SSFP sequence used for the second excitation, and the Balance-SSFP sequence used for the second excitation and the Balance-SSFP sequence used for the first excitation constitute a double-excitation Balance-SSFP sequence. The difference from the aforementioned sequence is that two positive-phase RF pulses α+ are applied between TR. In this embodiment, the image reconstructed by the magnetic resonance signal collected in the TE period is fused with the image reconstructed by the magnetic resonance signal collected in the first excitation to obtain the label data of the deep learning neural network.

[0068] In one embodiment, an implementation method of "obtaining a difference image between a low signal-to-noise ratio image and a high signal-to-noise ratio image of the detection object" in the above step S101 includes: subtracting the low signal-to-noise ratio image from the high signal-to-noise ratio image to obtain a difference image.

[0069] In this embodiment, on the basis of obtaining the low signal-to-noise ratio image and the high signal-to-noise ratio image, the computer device obtains the difference image between the low signal-to-noise ratio image and the high signal-to-noise ratio image. Specifically, the computer device directly subtracts the low signal-to-noise ratio image from the high signal-to-noise ratio image to obtain the difference image. For example, the pixel value of each pixel of the low signal-to-noise ratio image is subtracted from the pixel value of the pixel of the high signal-to-noise ratio image. Of course, the computer device can also obtain the difference image between the low signal-to-noise ratio image and the high signal-to-noise ratio image by other methods, which are not limited thereto.

[0070] The image denoising method provided in this embodiment performs denoising on a low signal-to-noise ratio image of a detection object to obtain a high signal-to-noise ratio image of the detection object, and then subtracts the low signal-to-noise ratio image from the high signal-to-noise ratio image to obtain a difference image.

[0071] Before extracting the additional information image from the above difference image, it is necessary to first determine whether there is non-noise information in the difference image before executing the step of extracting the additional information image. Figure 5 As shown, in one embodiment, the method further includes:

[0072] S201, detecting whether there is non-noise information in the difference image.

[0073] The computer device detects whether there is non-noise information in the difference image by identifying the information in the difference image, or detects the signal distribution in the difference image through a signal distribution algorithm to determine whether there is additional information based on the signal distribution.

[0074] S202: If non-noise information exists, a step of extracting an additional information image from the difference image is performed.

[0075] After determining that there is non-noise information in the difference image in the above steps, the computer device starts to execute the step of extracting the additional information image from the difference image. In this way, if it is found that there is no non-noise information in the difference image, there is no need to execute the step of extracting the additional information image, thereby saving resources and improving image processing efficiency.

[0076] Optionally, taking the detection of signal distribution in the difference image as an example, Figure 6 As shown, one implementation method of detecting whether there is non-noise information in the difference image includes:

[0077] S301, obtaining signal distribution in a difference image.

[0078] Under normal circumstances, the signal distribution in the image is all noise without other information, but if there is non-noise information other than noise, it will affect the distribution of noise in the image. Therefore, after the computer device detects the signal distribution in the difference image, it obtains the signal distribution and then analyzes the signal distribution to see if there is non-noise information other than noise. For example, it analyzes whether the texture of the noise distribution is symmetrical or smooth.

[0079] S302: If the signal distribution is all noise, determine that there is no non-noise information in the difference image; if the signal distribution contains information other than noise, determine that there is non-noise information in the difference image.

[0080] If the analysis result is that the signal distribution is all noise, then the computer device determines that there is no non-noise information in the difference image. Otherwise, if the analysis result is that the signal distribution contains information other than noise, then it is determined that there is non-noise information in the difference image. This embodiment can increase the accuracy of detecting whether there is non-noise information by judging whether there is non-noise information in the difference image based on the signal distribution in the difference image. It can be understood that the accuracy of detecting whether there is non-noise information in the difference image helps to extract the accuracy of the additional information image from the difference image later.

[0081] In some scenarios, in order to further increase the accuracy of obtaining the additional information image and the accuracy of correcting the image, the above processing can be performed on the image in different regions by windowing. Figure 7 As shown, the step S101 above acquires a difference image between a low signal-to-noise ratio image and a high signal-to-noise ratio image of the detection object, and further includes:

[0082] S401, dividing the detection object into multiple sub-areas.

[0083] When performing region processing on an image, the computer device may first divide the detection object (for example, a pre-scanned image or an overview image of the detection object) into a plurality of sub-regions. For example, after presetting the size of the sub-region, the computer device divides the detection object into a plurality of sub-regions according to the actual size of the detection object combined with the preset size of the sub-region. Alternatively, there is no need to preset the size, and a set of region division rules is set. The computer device automatically determines the size of the sub-region to be divided according to the rule and the actual size of the detection object to be divided.

[0084] S402, obtaining a difference image between a low signal-to-noise ratio image and a high signal-to-noise ratio image of each sub-region.

[0085] Taking the divided sub-regions as units, the computer device obtains the low signal-to-noise ratio image and the high signal-to-noise ratio image of each sub-region, wherein the method of obtaining the difference image can refer to the description in the above embodiment and will not be repeated here.

[0086] In this embodiment, the detection object is divided into multiple sub-areas, and then a difference image between a low signal-to-noise ratio image and a high signal-to-noise ratio image of each sub-area is obtained, and the sub-areas are used as units, so that the difference image can be obtained more quickly and accurately.

[0087] Accordingly, after the detection object is divided into multiple sub-regions, each sub-region can be further used as a unit to extract non-noise information on the difference image of each sub-region. In one embodiment, the method also includes: extracting the non-noise information in the difference image of each sub-region, and correcting the high signal-to-noise ratio image of each sub-region based on the extracted non-noise information to obtain a corrected high signal-to-noise ratio image of each sub-region.

[0088] Based on the difference image of each sub-region, the corresponding non-noise information is extracted from each difference image, wherein the method of extracting the non-noise information can also be referred to the description in the above embodiment, which will not be repeated here. After extracting the additional information, each sub-region is still used as a unit, and each sub-region is corrected according to the non-noise information to obtain a corrected sub-region image, and finally the corrected sub-region images are fused to obtain a corrected high signal-to-noise ratio and high-quality detection object. The advantage of dividing the detection object into multiple sub-regions and performing the above-mentioned processing on the sub-region as a unit is that many sub-regions can be processed in parallel, thereby improving the processing efficiency.

[0089] like Figure 8 Shown and Fig. 9 As shown, the present application also provides an embodiment of an image noise reduction method, which includes:

[0090] S1, obtain a low signal-to-noise ratio image of the detection object, such as Fig. 9 (a)

[0091] S2, denoising the low signal-to-noise ratio image of the detection object to obtain a high signal-to-noise ratio image of the detection object, such as Fig. 9 (b)

[0092] S3, obtaining a difference image between the low signal-to-noise ratio image and the high signal-to-noise ratio image, such as Fig. 9 (c)

[0093] S4, obtaining signal distribution in the difference image;

[0094] S5, if the signal distribution is all noise, it is determined that there is no non-noise information in the difference image; if the noise distribution includes information other than the noise, it is determined that there is non-noise information in the difference image;

[0095] S6, if there is non-noise information, extract the additional information image from the difference image, such as Fig. 9 (d)

[0096] S7, correcting the high signal-to-noise ratio image of the detection object according to the non-noise information in the additional information image, such as Fig. 9 (e) in .

[0097] This embodiment corrects an image that has a high signal-to-noise ratio but contains some erroneous information (non-noise information). The obtained corrected image retains the characteristics of the high signal-to-noise ratio while removing the non-noise information, so that the structure in the image retains the original real features, thereby improving the image quality.

[0098] The implementation principles and technical effects of each step in the image denoising method provided in this embodiment are similar to those in the previous embodiments of the image denoising method, and will not be repeated here. Figure 8 The implementation method of each step in the embodiment is only an example, and the implementation method is not limited. The order of each step can be adjusted in actual application as long as the purpose of each step can be achieved.

[0099] On the basis of the noise reduction method involved in the above-mentioned embodiment, the present application further proposes a method for processing medical image data, where the medical image data is acquired by using a planar echo sequence. The method may include:

[0100] First, K-space data is acquired, and the K-space data includes odd-numbered line echo data and even-numbered line echo data. Optionally, the K-space data is acquired by collecting multiple channels after excitation of a planar echo sequence, and the K-space data includes odd-numbered line echo data and even-numbered line echo data.

[0101] Secondly, the odd-numbered lines of echo data of the K-space data are used to form the first K-space data, and the even-numbered lines of echo data of the K-space data are used to form the second K-space data.

[0102] Again, the first K space data and the second K space data are reconstructed respectively to obtain a first reconstructed image and a second reconstructed image. Optionally, the reconstruction algorithm may be, for example, a compressed sensing algorithm, a parallel imaging algorithm, etc. It is understandable that the K space where the first K space data and the second K space data are located is an undersampled K space. Before reconstructing the first reconstructed image and the second reconstructed image, data recovery may also be performed on the two K spaces corresponding to the first K space data and the second K space data.

[0103] Next, the first reconstructed image is subjected to denoising to obtain a first high signal-to-noise ratio image; the second reconstructed image is subjected to denoising to obtain a second high signal-to-noise ratio image; and the first reconstructed image and the first high signal-to-noise ratio image are subjected to the following Figure 2 or Figure 8 The method flow shown in , obtains the corrected first high signal-to-noise ratio image; similarly, for the second reconstructed image and the second high signal-to-noise ratio image, the following is used Figure 2 or Figure 8 According to the method flow shown in , a corrected second image with the highest signal-to-noise ratio is obtained.

[0104] Finally, the relative phase difference between adjacent echo data is determined according to the corrected first high signal-to-noise ratio image and the corrected second high signal-to-noise ratio image; and the K-space data is corrected according to the relative phase difference to obtain the corrected K-space data.

[0105] In this embodiment, determining the relative phase difference of adjacent echo data according to the corrected first high signal-to-noise ratio image and the corrected second high signal-to-noise ratio image includes: obtaining the phase value of the odd-numbered rows of echo data according to the corrected first high signal-to-noise ratio image; obtaining the phase value of the even-numbered rows of echo data according to the corrected second high signal-to-noise ratio image; and obtaining the relative phase difference between the odd-numbered rows of echo data and the even-numbered rows of echo data according to the obtained phase values ​​of the odd-numbered rows of echo data and the obtained phase values ​​of the even-numbered rows of echo data.

[0106] In this embodiment, correcting the K-space data according to the relative phase difference to obtain the corrected K-space data may include: correcting the odd-numbered row echo data of the K-space data according to the relative phase difference to obtain the corrected odd-numbered row echo data, and merging the corrected odd-numbered row echo data with the even-numbered row echo data of the K-space data to obtain the corrected K-space data; or, correcting the even-numbered row echo data of the K-space data according to the relative phase difference to obtain the corrected even-numbered row echo data, and merging the corrected even-numbered row echo data with the odd-numbered row echo data of the K-space data to obtain the corrected K-space data. In the embodiment of the present application, the medical image is first subjected to noise reduction processing to obtain a corrected second high signal-to-noise ratio image, and then the relative phase difference of the adjacent row echo data is calculated according to the corrected high signal-to-noise ratio image corresponding to the odd-numbered row echo data and the even-numbered row echo data, and then the K-space data is corrected according to the phase difference. The phase correction of adjacent echo signals is performed using the medical image of the high signal-to-noise ratio image, thereby improving the accuracy of signal correction.

[0107] It should be understood that although Figure 2-9 The steps in the flowchart are shown in sequence as indicated by the arrows, but these steps are not necessarily executed in the order indicated by the arrows. Unless otherwise specified in this document, there is no strict order restriction for the execution of these steps, and these steps can be executed in other orders. Moreover, Figure 2-9 At least part of the steps may include multiple steps or multiple stages. These steps or stages are not necessarily performed at the same time, but can be performed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed in turn or alternately with other steps or at least part of the steps or stages in other steps.

[0108] In one embodiment, Fig.10 As shown, an image noise reduction device is provided, the device comprising:

[0109] The acquisition module 10 is used to acquire a low signal-to-noise ratio image and a high signal-to-noise ratio image of the detection object.

[0110] A difference finding module 11 is used to obtain a difference image between a low signal-to-noise ratio image and a high signal-to-noise ratio image of the detection object;

[0111] An extraction module 12 is used to extract an additional information image from the difference image; the additional information image includes non-noise information in the difference image;

[0112] The correction module 13 is used to correct the high signal-to-noise ratio image of the detection object according to the non-noise information to obtain a corrected high signal-to-noise ratio image.

[0113] In one embodiment, the acquisition module 10 is specifically used to perform noise reduction on a low signal-to-noise ratio image of the detection object through a preset deep learning noise reduction network to obtain a high signal-to-noise ratio image of the detection object.

[0114] In one embodiment, the difference module 11 is specifically used to subtract the image with a low signal-to-noise ratio from the image with a high signal-to-noise ratio to obtain a difference image.

[0115] In one embodiment, the apparatus further comprises:

[0116] The detection module is used to detect whether there is non-noise information in the difference image; if there is non-noise information, the step of extracting the additional information image from the difference image is executed.

[0117] In one embodiment, the detection module includes:

[0118] A signal acquisition unit, used for acquiring signal distribution in the difference image;

[0119] The information determination unit is used to determine that there is no non-noise information in the difference image if the signal distribution all appears as noise; and to determine that there is non-noise information in the difference image if the signal distribution contains information other than noise.

[0120] In one embodiment, the extraction module 11 includes:

[0121] A division unit, used for dividing the detection object into a plurality of sub-areas;

[0122] The acquisition unit is used to acquire a difference image between the low signal-to-noise ratio image and the high signal-to-noise ratio image of each sub-region.

[0123] In one embodiment, the device further includes: a processing module for extracting non-noise information from the difference image of each sub-region, and correcting the high signal-to-noise ratio image of each sub-region based on the extracted non-noise information to obtain a corrected high signal-to-noise ratio image of each sub-region.

[0124] The implementation principles and technical effects of all the image denoising devices provided in the above embodiments are similar to those of the above-mentioned image denoising method embodiments, and are not described in detail here. Each module in the above-mentioned image denoising device can be implemented in whole or in part by software, hardware, and a combination thereof. Each of the above-mentioned modules can be embedded in or independent of a processor in a computer device in the form of hardware, or can be stored in a memory in a computer device in the form of software, so that the processor can call and execute operations corresponding to each of the above modules.

[0125] In one embodiment, a computer device is provided. The computer device may be a terminal, and its internal structure diagram may be as described above. Figure 1As shown. The computer device includes a processor, a memory, a communication interface, a display screen and an input device connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be achieved through WIFI, an operator network, NFC (near field communication) or other technologies. When the computer program is executed by the processor, an image noise reduction method is implemented. The display screen of the computer device can be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device can be a touch layer covered on the display screen, or a button, trackball or touchpad set on the computer device housing, or an external keyboard, touchpad or mouse, etc.

[0126] Those skilled in the art will appreciate that the above Figure 1 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0127] In one embodiment, a computer device is provided, including a memory and a processor, wherein a computer program is stored in the memory, and when the processor executes the computer program, the following steps are implemented:

[0128] Acquire low signal-to-noise ratio images and high signal-to-noise ratio images of the detection object;

[0129] Obtaining a difference image between a low signal-to-noise ratio image and a high signal-to-noise ratio image of the detection object;

[0130] Extracting an additional information image from the difference image; the additional information image includes non-noise information in the difference image;

[0131] According to the non-noise information, the high signal-to-noise ratio image of the detection object is corrected to obtain a corrected high signal-to-noise ratio image.

[0132] The above embodiment provides a computer device, whose implementation principle and technical effect are similar to those of the above method embodiment, and will not be repeated here.

[0133] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored, and when the computer program is executed by a processor, the following steps are implemented:

[0134] Acquire low signal-to-noise ratio images and high signal-to-noise ratio images of the detection object;

[0135] Obtaining a difference image between a low signal-to-noise ratio image and a high signal-to-noise ratio image of the detection object;

[0136] Extracting an additional information image from the difference image; the additional information image includes non-noise information in the difference image;

[0137] According to the non-noise information, the high signal-to-noise ratio image of the detection object is corrected to obtain a corrected high signal-to-noise ratio image.

[0138] The above embodiment provides a computer-readable storage medium, whose implementation principle and technical effect are similar to those of the above method embodiment, and will not be repeated here.

[0139] Those of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory or optical memory, etc. Volatile memory can include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM).

[0140] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, 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 specification.

[0141] The above-mentioned embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the invention patent. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the patent of the present application shall be subject to the attached claims.

Claims

1. An image denoising method, characterized in that: The method comprises: Acquire low signal-to-noise ratio images and high signal-to-noise ratio images of the detection object; Acquire a difference image between a low signal-to-noise ratio image and a high signal-to-noise ratio image of the detection object; the difference image is obtained by subtracting the low signal-to-noise ratio image from the high signal-to-noise ratio image; Extracting an additional information image from the difference image; the additional information image includes non-noise information in the difference image; the additional information includes error information other than noise; The erroneous information in the high signal-to-noise ratio image of the detection object is removed to obtain a corrected high signal-to-noise ratio image.

2. The method according to claim 1, characterized in that The step of acquiring a low signal-to-noise ratio image and a high signal-to-noise ratio image of the detection object comprises: The low signal-to-noise ratio image of the detection object is denoised by a preset deep learning denoising network to obtain a high signal-to-noise ratio image of the detection object.

3. The method according to claim 1 or 2, characterized in that: Before extracting the additional information image from the difference image, the method further comprises: Detecting whether the non-noise information exists in the difference image; If the non-noise information exists, the step of extracting the additional information image from the difference image is performed.

4. The method according to claim 3, characterized in that The detecting whether the non-noise information exists in the difference image includes: Obtaining signal distribution in the difference image; If the signal distributions all appear as noise, determining that the non-noise information does not exist in the difference image; If the signal distribution contains information other than noise, it is determined that the non-noise information exists in the difference image.

5. The method according to claim 1 or 2, characterized in that: The step of obtaining a difference image between a low signal-to-noise ratio image and a high signal-to-noise ratio image of the detection object further includes: Dividing the detection object into a plurality of sub-areas; A difference image between the low signal-to-noise ratio image and the high signal-to-noise ratio image of each of the sub-regions is obtained.

6. The method according to claim 5, characterized in that The method further comprises: Non-noise information is extracted from the difference image of each of the sub-regions, and the high signal-to-noise ratio image of each of the sub-regions is corrected according to the extracted non-noise information to obtain a corrected high signal-to-noise ratio image of each of the sub-regions.

7. An image noise reduction device, characterized in that: The device comprises: An acquisition module, used for acquiring a low signal-to-noise ratio image and a high signal-to-noise ratio image of the detection object; A difference-finding module, used for obtaining a difference image between a low signal-to-noise ratio image and a high signal-to-noise ratio image of the detection object; the difference image is obtained by subtracting the low signal-to-noise ratio image from the high signal-to-noise ratio image; An extraction module, used for extracting an additional information image from the difference image; the additional information image includes non-noise information in the difference image; the additional information includes error information other than noise; The correction module is used to remove the erroneous information in the high signal-to-noise ratio image of the detection object according to the non-noise information to obtain a corrected high signal-to-noise ratio image.

8. The device according to claim 7, characterized in that The acquisition module is also used for: The low signal-to-noise ratio image of the detection object is denoised by a preset deep learning denoising network to obtain a high signal-to-noise ratio image of the detection object.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 6 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

Images