Image processing methods, image processing apparatus and storage media

Abstract

An image processing method, an image processing apparatus, and a storage medium are disclosed. The image processing method includes: acquiring an initial imaging image including a target object, obtained based on a reconstruction processing algorithm; acquiring a first calculated value of a scaling parameter and a second calculated value of a translation parameter; preprocessing the initial imaging image using the first and second calculated values ​​to obtain a processed image; and using the processed image for image segmentation processing based on an image segmentation algorithm to obtain a segmented image for the target object. The first and second calculated values ​​correspond to the reconstruction processing algorithm and the image segmentation algorithm, respectively. In the preprocessing, the pixel count of the processed image is equal to the pixel count of the initial imaging image multiplied by the first calculated value, which is then increased by the second calculated value. This image processing method overcomes the limitations of using a single image segmentation algorithm and improves the reusability of image segmentation algorithms.

Description

Technical Field

[0001] Embodiments of this disclosure relate to an image processing method, an image processing apparatus, and a storage medium. Background Technology

[0002] In today's medical diagnostic field, the most widely used imaging technique for examining potential lesions in patients is CT (Computed Topography). CT imaging is a technique that uses precisely collimated X-ray beams, gamma rays, etc., to obtain a series of projection data about a target from multiple observation angles, and then obtains a tomographic image of the target through image reconstruction. It has the characteristics of fast scanning time and clear images, and can be used to examine a variety of common diseases (such as tumors).

[0003] In recent years, with the development of AI technology, some deep learning-based image segmentation algorithms have been applied to the medical diagnosis process. These algorithms can segment and extract specific human organs or lesion areas from medical images for subsequent manual or AI-based medical diagnosis. For example, the nnU-Net algorithm, which is widely used in the field of medical image segmentation, can accurately segment human organs or tissues such as the heart, spleen, and hippocampus, playing an important role in the medical diagnosis process. Summary of the Invention

[0004] At least one embodiment of this disclosure provides an image processing method, which includes: acquiring an initial imaging image including a target object obtained based on a reconstruction processing algorithm; acquiring a first calculated value of a scaling parameter and a second calculated value of a translation parameter; preprocessing the initial imaging image using the first calculated value and the second calculated value to obtain a processed image; and using the processed image for image segmentation processing based on an image segmentation algorithm to obtain a segmented image for the target object, wherein the first calculated value and the second calculated value correspond to the reconstruction processing algorithm and the image segmentation algorithm, and in the preprocessing, the pixels of the processed image = the pixels of the initial imaging image * the first calculated value + the second calculated value.

[0005] At least one embodiment of this disclosure provides an image processing apparatus, including: a processing unit and a memory, wherein one or more computer program modules are stored in the memory; wherein the one or more computer program modules are configured to implement the image processing method described above when executed by the processing unit.

[0006] At least one embodiment of this disclosure provides a non-transitory readable storage medium, wherein computer instructions are stored on the non-transitory readable storage medium, wherein the computer instructions, when executed by a processor, implement the image processing method described above. Attached Figure Description

[0007] To more clearly illustrate the technical solutions of the embodiments of this disclosure, the accompanying drawings of the embodiments will be briefly described below. Obviously, the drawings described below only relate to some embodiments of this disclosure and are not intended to limit this disclosure.

[0008] Figure 1A and Figure 1B A schematic diagram illustrating the principle of the direct back projection method is shown.

[0009] Figure 2 A schematic diagram illustrating the principle of the filtered back projection method is shown.

[0010] Figure 3 A flowchart of a deep learning algorithm for brain tumor segmentation is presented.

[0011] Figure 4 A schematic diagram comparing an embodiment of this disclosure with a prior art solution is shown;

[0012] Figure 5 A schematic diagram illustrating at least one embodiment of the present disclosure provides an image processing method;

[0013] Figure 6 This diagram illustrates an example of the process of obtaining parameter values ​​for preprocessing.

[0014] Figure 7 A schematic block diagram of an electronic device provided for at least one embodiment of this disclosure. Detailed Implementation

[0015] To make the objectives, technical solutions, and advantages of the embodiments of this disclosure clearer, the technical solutions of the embodiments of this disclosure will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this disclosure. All other embodiments obtained by those skilled in the art based on the described embodiments of this disclosure without creative effort are within the scope of protection of this disclosure.

[0016] Unless otherwise defined, the technical or scientific terms used herein should have the ordinary meaning understood by one of ordinary skill in the art to which this disclosure pertains. The terms “first,” “second,” and similar terms used in this disclosure do not indicate any order, quantity, or importance, but are merely used to distinguish different components. Similarly, terms such as “comprising” or “including” mean that the element or object preceding the word encompasses the elements or objects listed following the word and their equivalents, without excluding other elements or objects. Terms such as “connected” or “linked” are not limited to physical or mechanical connections, but can include electrical connections, whether direct or indirect. Terms such as “upper,” “lower,” “left,” and “right” are used only to indicate relative positional relationships, which may change accordingly when the absolute position of the described objects changes.

[0017] In CT technology, various reconstruction algorithms can be used to obtain tomographic images from projection data acquired by CT. These reconstruction algorithms include Direct Back Projection (DBP) and Filtered Back Projection (FBP), which will be briefly introduced below.

[0018] Direct Back Projection (DBP) is a reconstruction algorithm that backprojects the measured projection data "back along the original path" onto each pixel of the projection line. The value of a pixel in the tomographic plane is considered as the sum (or average) of the projection values ​​of all rays passing through that pixel. In polar coordinates, the back projection expression for the continuous case is:

[0019]

[0020] Among them, angle The position of the projection line can be represented by [reference needed]. Figure 1A and Figure 1B As shown, for example, see Chapter 2, Sections 2.1 and 2.2 of “Fan Huiyun. Research on CT Image Filtering Back Projection Reconstruction Algorithm [D]. Northwestern Polytechnical University, 2007”.

[0021] The drawback of the direct backprojection reconstruction algorithm is the introduction of star-shaped artifacts, meaning that points with zero density in the original image may not have zero density after reconstruction, causing image distortion.

[0022] Filtered Back Projection (FBP) involves first correcting (filtering) the projection data, then performing a backprojection operation on the corrected projection data to obtain a reconstructed image that is approximately distortion-free. Mathematically, the characteristic of FBP is that it performs convolution processing on the projection at each acquired projection angle before backprojection; the one-dimensional Fourier transform of the projection is equivalent to performing a two-dimensional Fourier transform on the original image. Figure 2 The principle of the filtered back projection method is shown, for example, see Chapter 2.3 of "Fan Huiyun. Research on CT Image Filtered Back Projection Reconstruction Algorithm [D]. Northwestern Polytechnical University, 2007".

[0023] For small structures such as tumors, methods such as thresholding, region growing, atlas segmentation, variable model segmentation, and machine learning can be used. With the improvement of hardware computing resources and the increase in brain tumor imaging data, faster and more accurate deep learning methods are gradually gaining attention. Deep learning methods typically use Convolutional Neural Networks (CNNs) as their core, undergoing iterative steps such as training, testing, and validation to ultimately achieve image segmentation results with extremely high accuracy for specific application scenarios.

[0024] For example, Figure 3 A flowchart illustrating an exemplary deep learning algorithm for brain tumor segmentation is provided. The flowchart describes data preprocessing, including brain region standardization, offset field removal, skull removal, and registration. This data can be used as training and validation sets, or as data for testing or classification. The neural networks employed include multiple convolutional neural networks (CNNs). The CNNs are iteratively trained using the training and validation sets, and the model is evaluated and saved. The trained CNN model can then be used for prediction and to obtain prediction results.

[0025] As mentioned earlier, segmenting the reconstructed image is a necessary means of processing the data from the aforementioned instruments. However, the medical image segmentation algorithms described above have the problem of being sensitive to pixel values ​​but insensitive to shapes. For example, a well-trained medical image segmentation algorithm can only identify tumors with a pixel value of 0 and a specific shape. If given an image with a tumor region having a pixel value of 150, even if the shape is very similar to a tumor, the image segmentation algorithm cannot identify it. However, the tumor pixel values ​​in medical images obtained from different image reconstruction algorithms vary. Therefore, this can lead to an AI medical image segmentation algorithm trained on images reconstructed by one reconstruction algorithm as a dataset, which may not be able to effectively segment images reconstructed by another reconstruction algorithm.

[0026] For example, for a CT reconstruction algorithm that uses the Fourier central slice theorem, the pixel value of the tumor region is about 150, while for a back projection (DBP) CT reconstruction algorithm, the pixel value of the tumor region is about 0. The difference in the pixel value of the tumor region obtained by the two CT reconstruction algorithms is large, and they are not suitable for the same AI medical image segmentation algorithm.

[0027] To address the above problems, embodiments of this disclosure propose an image processing method and corresponding image processing apparatus that enables image segmentation algorithms to be applied to reconstructed images obtained from different reconstruction algorithms. This method performs a simple preprocessing on the reconstructed image, making the pixel values ​​corresponding to the target object (target region) close to, for example, the tumor pixel feature values ​​that a trained segmentation algorithm can normally identify. The preprocessing method provided in this embodiment includes: performing a linear transformation on all pixels of the image, i.e., multiplying by a fixed coefficient (scaling parameter) and then adding (or subtracting) a fixed value (translation parameter). This preprocessing requires relatively low computational overhead and can overcome the application limitations of a single image segmentation algorithm, greatly improving the reusability of image segmentation algorithms.

[0028] Figure 4 A schematic diagram comparing an embodiment of this disclosure with a prior art solution is shown. In the prior art solution, the image segmentation algorithm is only applicable to reconstruction algorithm 1 (where the pixel value of the target object region is 0), and not to reconstruction algorithm 2 (where the pixel value of the target object region is 150). However, the image processing method of the embodiment of this disclosure performs preprocessing corresponding to reconstruction algorithm 1 on the reconstructed image to obtain a preprocessed image, which is then applicable to the image segmentation algorithm used. Furthermore, the reconstructed image of reconstruction algorithm 2 also undergoes preprocessing corresponding to reconstruction algorithm 2 to obtain a preprocessed image, which is also applicable to the image segmentation algorithm used. Therefore, image segmentation can be successfully achieved for both reconstruction algorithm 1 and reconstruction algorithm 2.

[0029] Figure 5 A schematic diagram illustrating at least one embodiment of this disclosure provides an image processing method. For example... Figure 5 As shown, at least one embodiment of this disclosure provides an image processing method including the following steps 101 to 104:

[0030] Step 101: Obtain the initial imaging image of the target object based on the reconstruction processing algorithm.

[0031] Step 102: Obtain the first calculated value A of the scaling parameter and the second calculated value B of the translation parameter.

[0032] Step 103: Preprocess the initial imaging image using the first calculated value A and the second calculated value B to obtain the processed image.

[0033] Step 104: Use the processed image for image segmentation processing based on the image segmentation algorithm to obtain a segmented image of the target object.

[0034] In the above method, the first calculated value A of the scaling parameter and the second calculated value B of the translation parameter correspond to the above reconstruction processing algorithm and the above image segmentation algorithm. In the above preprocessing, the pixel value of the processed image = the pixel value of the initial image * the first calculated value A + the second calculated value B, that is, scaling transformation and translation transformation are performed pixel by pixel.

[0035] For step 101, for example, a reconstruction processing algorithm is used to process the image acquisition data obtained from the image acquisition device to obtain an initial imaging image. For example, the image acquisition device includes: a computed tomography (CT) scanner, a magnetic resonance imaging (MRI) device, an ultrasound imaging device, etc., for example, these image acquisition devices are medical image acquisition devices, and the embodiments of this disclosure are not limited thereto. The target object mentioned above can be a tumor, heart, spleen, hippocampus, vegetative nodule, flaw detection area, etc. The digital medical image obtained by the image acquisition device is, for example, in DICOM (Digital Imaging and Communications in Medicine) format, where pixels record grayscale information.

[0036] For example, in the embodiments of this disclosure, the reconstruction processing algorithms include: direct back projection, filtered back projection, convolutional back projection, differential-Hilbert back projection, gradient descent, ML-EM iterative reconstruction, etc., and the embodiments of this disclosure do not limit these algorithms.

[0037] For step 104, for example, the image segmentation algorithm includes: threshold segmentation, region growing, spectral segmentation, variable model segmentation, deep learning-based image segmentation, etc.; for example, deep learning-based image segmentation algorithms may include the nnU-Net algorithm, etc.

[0038] For step 102, for example, in at least one example, obtaining the first calculated value A of the scaling parameter and the second calculated value B of the translation parameter may include steps 1021-1022:

[0039] Step 1021: Obtain the reference reconstructed image based on the reconstruction processing algorithm, and obtain the reference segmentation image that corresponds to the reference reconstructed image and is accurately segmented;

[0040] Step 1022: Use the reference reconstructed image and the reference segmented image to obtain the first calculated value A of the scaling parameter and the second calculated value B of the translation parameter.

[0041] In this disclosure, "accurate segmentation" refers to a segmentation accuracy greater than a certain threshold (e.g., 90% or 95%), not that the segmentation image is 100% accurate. A reference reconstructed image and a corresponding accurate segmentation reference image are prepared in advance. This reference segmentation image can be obtained using the image segmentation algorithm used in step 104 above, or it can be obtained using another image segmentation algorithm. This combination of "reference reconstructed image - reference segmentation image" yields a first calculated value A and a second calculated value B corresponding to the scaling parameter and translation parameter in the "reconstruction algorithm + segmentation algorithm" combination described above. For example, different combinations of "reconstruction algorithm + segmentation algorithm" can yield different combinations of "first calculated value A and second calculated value B".

[0042] For step 1022, in at least one example, the first calculated value A of the scaling parameter and the second calculated value B of the translation parameter are obtained using the reference reconstructed image and the reference segmentation image, including the following steps M1 to M4, and in each of the following steps, the first parameter is the scaling parameter and the corresponding second parameter is the translation parameter, or the first parameter is the translation parameter and the corresponding second parameter is the scaling parameter. The following explanation uses the example of the first parameter being the scaling parameter and the corresponding second parameter being the translation parameter:

[0043] Step M1: Set the first initial value A0 of the first parameter, the second initial value B0 of the second parameter, and the initial segmentation coefficient value D0.

[0044] Step M2: Fix the first parameter to the first initial value A0. Within the second value range corresponding to the second parameter B, starting from the second initial value B0, select the target value of the second parameter. Based on the target value of the second parameter and the initial segmentation coefficient D0, make a judgment to obtain the second current best value Bm and the intermediate segmentation coefficient value D1 of the second parameter.

[0045] Step M3: Fix the second parameter to the second current best value Bm, select the target value of the first parameter within the first value range corresponding to the first parameter A, and evaluate based on the target value of the first parameter and the intermediate segmentation coefficient value D1 to obtain the first current best value Am of the first parameter.

[0046] Step M4: Use the first current best value Am and the second current best value Bm as the calculated value A of the first parameter and the calculated value B of the second parameter to transform the initial imaging image.

[0047] For example, in step M3 above, selecting the target value of the first parameter includes: traversing or exhaustively searching within a first value range with a set first step length; for example, starting from a first initial value A0, selecting using the adopted method (e.g., traversal or exhaustive search). Alternatively, in step M2 above, selecting the target value of the second parameter includes: traversing or exhaustively searching within a second value range with a set second step length; for example, starting from a second initial value B0, selecting using the adopted method (e.g., traversal or exhaustive search). Here, the size of the first step length can be determined based on the size of the first value range; if it is too large, a relatively more preferred value may be missed; if it is too small, the computational load will increase. Similarly, the size of the second step length can be determined based on the size of the second value range. The sizes of the first step length and the second step length are different from each other.

[0048] For example, in step M2 above, the evaluation is based on the target value of the second parameter and the initial segmentation coefficient D; or, in step M3 above, the evaluation is based on the target value of the first parameter and the intermediate segmentation coefficient value D1, including the following steps M21 to M24:

[0049] Step M21: Transform the reference reconstructed image using the first current value Ax of the scaling parameter and the second current value Bx of the translation parameter to obtain the intermediate imaging image;

[0050] Step M22: Use a segmentation algorithm to segment the intermediate imaging image to obtain an intermediate segmented image;

[0051] Step M23: Obtain intermediate segmentation coefficients based on the intermediate segmentation image and the reference segmentation image;

[0052] Step M24: In response to the intermediate segmentation coefficient being greater than the current segmentation coefficient (or "current best segmentation coefficient") Dx, assign the intermediate segmentation coefficient to the current segmentation coefficient Dx, retain the parameter value of the currently selected parameter type as the current best value, and then change the parameter value of the currently selected parameter type. Alternatively, in response to the intermediate segmentation coefficient being less than or equal to the current segmentation coefficient Dx, change the parameter value of the currently selected parameter type.

[0053] In steps M21 to M24 above, the first initial value A0, the second initial value B0, and the initial segmentation coefficient D0 are used as the first current value Ax of the scaling parameter, the second current value Bx of the translation parameter, and the current segmentation coefficient Dx during the first evaluation; and the pixels of the intermediate imaging image = the pixels of the reference reconstructed image * the first current value Ax + the second current value Bx.

[0054] For example, in step M22 above, the intermediate segmentation coefficients are calculated using the intermediate segmentation image and the reference segmentation image, including the following steps:

[0055] Obtain the region T corresponding to the target object in the reference segmentation image (i.e., the region of the target object obtained by accurate segmentation), obtain the region P corresponding to the target object in the intermediate segmentation image, and calculate the intermediate segmentation coefficients as follows:

[0056]

[0057] In this context, DICE is the intermediate segmentation coefficient, T∩P represents the number of pixels that overlap between regions T and P, and T+P represents the sum of the number of pixels in regions T and P. The quality of segmentation is judged by the value of the segmentation coefficient DICE. The DICE value ranges from 0 to 1; a higher value indicates a better segmentation effect, with higher accuracy and reliability.

[0058] Figure 6 The illustration shows a schematic diagram of the process of obtaining parameter values ​​(i.e., the offset coefficient and scaling coefficient mentioned above) for preprocessing in an example of an embodiment of the present disclosure. The example is illustrated by traversal, and the process is to first traverse to determine the offset coefficient and then traverse to determine the scaling coefficient. The embodiments of the present disclosure are not limited to this traversal method, nor are they limited to the processing order of the offset coefficient and scaling coefficient. Although the following process is for the selected reconstruction algorithm and segmentation algorithm, it is not limited to specific reconstruction algorithms and segmentation algorithms.

[0059] See Figure 6 First, initial coefficient values ​​are set: the initial value of the scaling coefficient A is set to 1 (i.e., A = 1), the initial value of the offset coefficient B is set to -100 (i.e., B = -100, the lower limit of its value range), and the initial value of the segmentation coefficient, i.e., the current segmentation coefficient (optimal segmentation coefficient) DICE_m, is set to 0.9 (i.e., DICE_m = 0.9). In subsequent operations, during the traversal of coefficients A and B, a step size of 5 and a value range of -100 and 100 are selected for coefficient B, and a step size of 0.02 and a value range of 0.8 to 1.2 are selected for coefficient A. Furthermore, whenever the calculated DICE value after segmentation exceeds DICE_m, the DICE_m value is updated to record the optimal segmentation effect value that can be achieved.

[0060] More specifically, such as Figure 6As shown, after setting the initial values, i.e., A = 1, B = -100, and DICE_m = 0.9, the current values ​​of coefficients A and B (i.e., 1 and -100) are used to transform the image to obtain a new image (i.e., the intermediate image) = A * original image (i.e., the reference reconstructed image) + B. The new image is then segmented using the targeted AI image segmentation algorithm to obtain an intermediate segmented image. The DICE index is calculated using the intermediate segmented image and the reference segmented image. If the calculated DICE value is less than or equal to the current best segmentation coefficient DICE_m, the value of coefficient B is incremented by a step of 5. Otherwise, the best segmentation coefficient DICE_m is set to the currently calculated DICE value, and the best value of coefficient B, B_m, is set to the current value of coefficient B. Then, the value of coefficient B is incremented by a step of 5. After that, it is determined whether the value of coefficient B is greater than the upper limit of the value range, 100. If not, the previous process is repeated using the new value of coefficient B and the current value of coefficient A until the value of coefficient B is greater than 100, in order to select the best value of coefficient B, B_m. After the value of coefficient B is greater than 100, the optimal value of coefficient B_m is obtained.

[0061] After the above process, the coefficient A is set to 0.8, that is, A = 0.8 (the lower limit of its value range). Then it is determined whether A is greater than 1.2 (that is, the upper limit of its value range). If it is, the operation is stopped, and the optimal value A_m of coefficient A and the optimal value B_m of coefficient B are obtained. Otherwise, the selection continues to obtain the optimal value A_m of coefficient A. In the process of continuing to select the optimal value A_m for coefficient A, the current values ​​of coefficient A and coefficient B (i.e., the optimal value B_m) are used to transform the image to obtain a new image (i.e., the intermediate image) = A * original image (i.e., the reference reconstructed image) + B_m. The new image is then segmented using the targeted AI image segmentation algorithm to obtain intermediate segmented images. The DICE index is calculated using the intermediate segmented images and the reference segmented images. If the calculated DICE value is less than or equal to the current optimal segmentation coefficient DICE_m, the value of coefficient A is incremented by 0.02. Otherwise, the optimal segmentation coefficient DICE_m is set to the calculated DICE value, and the optimal value A_m of coefficient A is set to the current value of coefficient A. Then, the value of coefficient A is incremented by 0.02. Afterward, it is determined whether the value of coefficient A is greater than the upper limit of its range, 1.2. If the value of coefficient A is greater than 1.2, the optimal value A_m of coefficient A is obtained as described above.

[0062] Finally, the optimal values ​​A_m of coefficient A and B_m of coefficient B are used for image preprocessing for targeted reconstruction and AI segmentation algorithms.

[0063] Some embodiments of this disclosure also provide a processing apparatus including a processing unit and a memory, wherein one or more computer program modules are stored in the memory; the one or more computer program modules are configured to implement the image processing method as described in any embodiment of this disclosure when executed by the processing unit.

[0064] Some embodiments of this disclosure also provide a non-transitory readable storage medium, wherein computer instructions are stored on the non-transitory readable storage medium, wherein the computer instructions, when executed by a processor, implement the image method as described in any embodiment of this disclosure.

[0065] For example, the processing unit can be a central processing unit (CPU), a digital signal processor (DSP), or the like. For example, the non-transiently readable storage medium can be implemented as a memory, such as volatile memory and / or non-volatile memory. In the above embodiments, the memory can be volatile memory, for example, it may include random access memory (RAM) and / or cache, etc. Non-volatile memory may include, for example, read-only memory (ROM), hard disk, erasable programmable read-only memory (EPROM), portable compact disc read-only memory (CD-ROM), USB memory, flash memory, etc.

[0066] The memory can also store various applications (code, instructions) and various data, as well as various data used and / or generated by the applications. In some examples, it can also store memory tag information (and memory tags) corresponding to the data or instructions.

[0067] Some embodiments of this disclosure also provide an electronic device that includes the computer device described above, which includes the processing device described above or is capable of executing the memory access method described above.

[0068] Figure 7 This is a schematic block diagram of an electronic device provided for at least one embodiment of the present disclosure. The electronic device in the embodiments of the present disclosure may include, but is not limited to, mobile terminals such as laptops, PDAs (personal digital assistants), and PADs (tablet computers), as well as fixed terminals such as desktop computers.

[0069] Figure 7 The illustrated electronic device 1000 is merely an example and should not be construed as limiting the functionality and scope of the embodiments disclosed herein. For example, such as Figure 7As shown, in some examples, electronic device 1000 includes an image processing apparatus according to embodiments of the present disclosure. This image processing apparatus can perform various appropriate actions and processes, such as the image processing method of embodiments of the present disclosure, based on a program stored in read-only memory (ROM) 1002 or a program loaded from storage device 1008 into random access memory (RAM) 1003. The RAM 1003 also stores various programs and data required for the operation of the computer system. Processor 1001, ROM 1002, and RAM 1003 are connected via bus 1004. Input / output (I / O) interface 1005 is also connected to bus 1004.

[0070] For example, the following components can be connected to I / O interface 1005: input devices 1006 including, for example, touch screens, touchpads, keyboards, mice, cameras, microphones, accelerometers, gyroscopes, etc.; output devices 1007 including, for example, liquid crystal displays (LCDs), speakers, vibrators, etc.; storage devices 1008 including, for example, magnetic tapes, hard disks, etc.; and communication devices 1009, such as network interface cards like LAN cards and modems, etc. Communication device 1009 allows electronic device 1000 to communicate wirelessly or wiredly with other devices to exchange data and perform communication processing via networks such as the Internet. Drive 1010 is also connected to I / O interface 1005 as needed. Removable storage media 1011, such as disks, optical disks, magneto-optical disks, semiconductor memories, etc., are installed on drive 1010 as needed so that computer programs read from them can be installed into storage device 1008 as needed.

[0071] Although Figure 7 An electronic device 1000 including various devices is shown; however, it should be understood that implementation or inclusion of all shown devices is not required. More or fewer devices may be implemented or included alternatively.

[0072] For example, the electronic device 1000 may further include a peripheral interface (not shown in the figure). This peripheral interface can be various types of interfaces, such as a USB interface, a Lightning interface, etc. The communication device 1009 can communicate wirelessly with a network and other devices, such as the Internet, an intranet, and / or a wireless network such as a cellular telephone network, a wireless local area network (LAN), and / or a metropolitan area network (MAN). Wireless communication can use any of a variety of communication standards, protocols, and technologies, including but not limited to Global System for Mobile Communications (GSM), Enhanced Data GSM Environment (EDGE), Wideband Code Division Multiple Access (W-CDMA), Code Division Multiple Access (CDMA), Time Division Multiple Access (TDMA), Bluetooth, Wi-Fi (e.g., based on IEEE 802.11a, IEEE 802.11b, IEEE 802.11g, and / or IEEE 802.11n standards), Voice over Internet Protocol (VoIP), Wi-MAX, protocols for email, instant messaging, and / or Short Message Service (SMS), or any other suitable communication protocol.

[0073] The following points should be noted regarding this disclosure:

[0074] (1) The accompanying drawings of the embodiments of this disclosure only involve the structures involved in the embodiments of this disclosure. Other structures can be referred to the general design.

[0075] (2) Where there is no conflict, the embodiments of this disclosure and the features in the embodiments can be combined with each other to obtain new embodiments.

[0076] The above are merely exemplary embodiments of this disclosure and are not intended to limit the scope of protection of this disclosure, which is determined by the appended claims.

Claims

1. An image processing method, comprising: Acquire an initial imaging image of the target object based on a reconstruction processing algorithm; Obtain the first calculated value of the scaling parameter and the second calculated value of the translation parameter; The initial image is preprocessed using the first calculated value and the second calculated value to obtain the processed image; The processed image is then used for image segmentation processing based on an image segmentation algorithm to obtain a segmented image for the target object. Wherein, the first calculated value and the second calculated value correspond to the reconstruction processing algorithm and the image segmentation algorithm, and in the preprocessing, the pixel of the processed image = the pixel of the initial imaging image * the first calculated value + the second calculated value; The acquisition of the first calculated value of the scaling parameter and the second calculated value of the translation parameter includes: Obtain a reference reconstructed image based on the reconstruction processing algorithm, and obtain a reference segmented image that corresponds to the reference reconstructed image and is accurately segmented; Using the reference reconstructed image and the reference segmented image, a traversal process is performed with the segmentation coefficient values ​​as the criterion to obtain the first calculated value of the scaling parameter and the second calculated value of the translation parameter.

2. The image processing method according to claim 1, wherein, Obtaining an initial imaging image including the target object based on the reconstruction processing algorithm includes: The reconstruction processing algorithm is used to process the image acquisition data obtained from the image acquisition device to obtain the initial imaging image.

3. The image processing method according to claim 2, wherein, The image acquisition device includes: a computed tomography (CT) device, a magnetic resonance imaging (MRI) device, or an ultrasound imaging device.

4. The image processing method according to claim 1, wherein, The reconstruction processing algorithms include: direct back projection, filtered back projection, convolutional back projection, differential-Hilbert back projection, gradient descent, and ML-EM iterative reconstruction.

5. The image processing method according to claim 1, wherein, The image segmentation algorithms include: threshold segmentation, region growing, spectral segmentation, variable model segmentation, or deep learning-based image segmentation.

6. The image processing method according to claim 1, wherein, Using the reference reconstructed image and the reference segmented image, a traversal process is performed with segmentation coefficient values ​​as the criterion to obtain the first calculated value of the scaling parameter and the second calculated value of the translation parameter, including: Set the first initial value of the first parameter, the second initial value of the second parameter, and the initial segmentation coefficient value; The first parameter is fixed at the first initial value. Within the second value range corresponding to the second parameter, starting from the second initial value, the target value of the second parameter is selected. Based on the target value of the second parameter and the initial segmentation coefficient, an evaluation is performed to obtain the second current best value and the intermediate segmentation coefficient value of the second parameter. The second parameter is fixed at the second current best value. Within the first value range corresponding to the first parameter, the target value of the first parameter is selected. Based on the target value of the first parameter and the intermediate segmentation coefficient value, a judgment is made to obtain the first current best value of the first parameter. The first current best value and the second current best value are used as the calculated values ​​of the first parameter and the second parameter, respectively, to transform the initial imaging image, wherein... The first parameter is the scaling parameter and the corresponding second parameter is the translation parameter; or, the first parameter is the translation parameter and the corresponding second parameter is the scaling parameter.

7. The image processing method according to claim 6, wherein, Selecting the target value of the first parameter includes: traversing or exhaustively searching within the range of the first value with the set first step length; Selecting the target value of the second parameter includes: traversing or exhaustively searching within the range of the second value with a set second step size.

8. The image processing method according to claim 6, wherein, Selecting a target value for the second parameter and evaluating it based on the initial segmentation coefficient, or selecting a target value for the first parameter and evaluating it based on the intermediate segmentation coefficient value, includes: The reference reconstructed image is transformed using the first current value of the scaling parameter and the second current value of the translation parameter to obtain an intermediate imaging image; The intermediate imaging image is segmented using the segmentation algorithm described above to obtain an intermediate segmented image; The intermediate segmentation coefficients are obtained based on the intermediate segmentation image and the reference segmentation image; If the intermediate segmentation coefficient is greater than the current segmentation coefficient, then the intermediate segmentation coefficient is assigned to the current segmentation coefficient, the parameter value of the currently selected parameter type is retained as the current optimal value, and then the parameter value of the currently selected parameter type is changed. Alternatively, if the intermediate segmentation coefficient is less than or equal to the current segmentation coefficient, then the parameter value of the currently selected parameter type is changed. Wherein, the first initial value, the second initial value, and the initial segmentation coefficient are used as the first current value of the scaling parameter, the second current value of the translation parameter, and the current segmentation coefficient during the first evaluation. The number of pixels in the intermediate image is equal to the number of pixels in the reference reconstructed image multiplied by the first current value and the second current value.

9. The image processing method according to claim 8, wherein, The intermediate segmentation coefficients are calculated using the intermediate segmentation image and the reference segmentation image, including: Obtain the region T corresponding to the target object in the reference segmentation image. Obtain the region P corresponding to the target object in the intermediate segmented image. The intermediate segmentation coefficient is calculated as follows: , Wherein, DICE is the intermediate segmentation coefficient. This indicates the number of pixels that overlap between regions T and P. This represents the sum of the number of pixels in region T and region P.

10. An image processing apparatus, comprising: Processing unit A memory on which one or more computer program modules are stored; The one or more computer program modules are configured to implement the image processing method as described in any one of claims 1-9 when executed by the processing unit.

11. A non-transitory readable storage medium, wherein, The non-transiently readable storage medium stores computer instructions, which, when executed by a processor, implement the image processing method as described in any one of claims 1-9.

Images