Lesion image determination method and device, electronic equipment and storage medium

Abstract

The application relates to the technical field of image processing, and particularly provides a lesion image determination method and device, electronic equipment and a storage medium, aiming to solve the problem of inaccurate lesion image determination and poor data compatibility. The method comprises the following steps: step 1, obtaining an apparent diffusion coefficient image based on a diffusion weighted image of a head of a target object; step 2, obtaining a non-lesion image and a first candidate lesion image based on the diffusion weighted image and the apparent diffusion coefficient image; step 3, obtaining a second candidate lesion image from the first candidate lesion image based on an apparent diffusion coefficient value and a lesion prediction probability value of each pixel point of the non-lesion image and the first candidate lesion image; step 4, if a change rate of the second candidate lesion image and a third candidate lesion image obtained through last iteration calculation exceeds a preset change rate, executing step 5; otherwise, determining the second candidate lesion image as a lesion image; and step 5, expanding the first candidate lesion image to a preset number of times of the current first candidate lesion image to execute step 3.

Description

Technical Field

[0001] This application relates to the field of image processing technology, specifically to a method, apparatus, electronic device, and storage medium for determining lesion images. Background Technology

[0002] DWI (Diffusion-Weighted Imaging) and ADC (Apparent Diffusion Coefficient) images are noisy, have many artifacts, and the appearance of lesions varies with the progression of the disease. Even for data in the standard range, the fixed thresholds used in academic research cannot adapt to differences in equipment and individual patient differences.

[0003] Moreover, the parameters of different devices vary greatly, and the image sequence values ​​given by the scanning equipment of different manufacturers are not standardized. The pixel values ​​of DWI and ADC are not necessarily within the standard range and may have undergone complex transformations. Even using conventional normalization such as z-normalization alone cannot transform them into a more consistent distribution, which causes great trouble for segmentation.

[0004] If the processed data is not used in the training process of the deep learning network, it may lead to incorrect segmentation results, unclear edge segmentation, or even missed detections. The accuracy of the output results of the deep learning network can only be corrected by adding a large amount of data to train the model. When the amount of data is insufficient and contains various discrepancies, it is equivalent to indirectly polluting the labeled data, which will lead to a decrease in inference accuracy. Summary of the Invention

[0005] This application addresses the shortcomings of the prior art by proposing a method, apparatus, electronic device, and storage medium for determining lesion images.

[0006] In a first aspect, embodiments of this application provide a method for determining lesion images, comprising: Step 1: obtaining an apparent diffusion coefficient image based on a diffusion-weighted image of the head of a target object, wherein the diffusion-weighted image and the apparent diffusion coefficient image have the same pixel matrix; Step 2: obtaining a lesion-free image and a first candidate lesion image based on the diffusion-weighted image and the apparent diffusion coefficient image; Step 3: obtaining a second candidate lesion image from the first candidate lesion image based on the apparent diffusion coefficient value and lesion prediction probability value corresponding to each pixel point of the lesion-free image and the first candidate lesion image; Step 4: if the rate of change between the second candidate lesion image and the third candidate lesion image obtained in the previous iteration exceeds a preset rate of change, then proceed to Step 5; otherwise, determine the second candidate lesion image as a lesion image; Step 5: enlarge the first candidate lesion image to a preset number of times its current size, and continue to execute Step 3.

[0007] Secondly, embodiments of this application provide a lesion image determination device, comprising: a first acquisition module, configured to acquire an apparent diffusion coefficient image based on a diffusion-weighted image of a target object's head, wherein the diffusion-weighted image and the apparent diffusion coefficient image have the same pixel matrix; a second acquisition module, configured to acquire a lesion-free image and a first candidate lesion image based on the diffusion-weighted image and the apparent diffusion coefficient image; a third acquisition module, configured to acquire a second candidate lesion image from the first candidate lesion image based on the apparent diffusion coefficient value and lesion prediction probability value corresponding to each pixel point of the lesion-free image and the first candidate lesion image; a lesion determination module, configured to, if the rate of change between the second candidate lesion image and the third candidate lesion image obtained in the previous iteration exceeds a preset rate of change, then jump to an image processing module; otherwise, determine the second candidate lesion image as a lesion image; and an image processing module, configured to expand the first candidate lesion image to a preset number of times its current size and jump to the third acquisition module.

[0008] Thirdly, embodiments of this application provide an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, it implements the method described in the first aspect above.

[0009] Fourthly, embodiments of this application provide a computer-readable storage medium storing a computer program that, when executed by a processor, implements the method described in the first aspect above.

[0010] The technical solutions provided in this application embodiment have at least the following technical effects or advantages:

[0011] The lesion image determination method of this application embodiment includes the following steps: Step 1: Obtaining an apparent diffusion coefficient image based on a diffusion-weighted image of the target object's head, wherein the diffusion-weighted image and the apparent diffusion coefficient image have the same pixel matrix; Step 2: Obtaining a lesion-free image and a first candidate lesion image based on the diffusion-weighted image and the apparent diffusion coefficient image; Step 3: Obtaining a second candidate lesion image from the first candidate lesion image based on the apparent diffusion coefficient value and lesion prediction probability value corresponding to each pixel point of the lesion-free image and the first candidate lesion image; Step 4: If the rate of change between the second candidate lesion image and the third candidate lesion image obtained in the previous iteration exceeds a preset rate of change, then proceed to Step 5; otherwise, the second candidate lesion image is determined to be a lesion image; Step 5: Expanding the first candidate lesion image to its current preset number of times and continuing to execute Step 3, which can improve the success rate of lesion determination under poor data compatibility, continuously correct the determination result of the candidate lesion image by iterative method, avoid the result error caused by using a single threshold, and improve the accuracy of lesion determination.

[0012] Additional aspects and advantages of this application will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of this application. Attached Figure Description

[0013] Various other advantages and benefits will become apparent to those skilled in the art upon reading the following detailed description of preferred embodiments. The accompanying drawings are for illustrative purposes only and are not intended to limit the scope of this application. Furthermore, the same reference numerals denote the same parts throughout the drawings. In the drawings:

[0014] Figure 1 A flowchart of a lesion image determination method provided in an embodiment of this application is shown;

[0015] Figure 2 This illustration shows a schematic diagram of a lesion image determination device provided in an embodiment of this application;

[0016] Figure 3 A schematic diagram of the structure of an electronic device provided in an embodiment of this application is shown. Detailed Implementation

[0017] Exemplary embodiments of this application will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of this application are shown in the drawings, it should be understood that this application may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided to enable a more thorough understanding of this application and to fully convey the scope of this application to those skilled in the art.

[0018] DWI and ADC images are noisy, have many artifacts, and the lesion manifestations vary with the progression of the disease. Even for data in the standard range, the fixed thresholds used in academic research cannot adapt to differences in equipment and individual patient differences.

[0019] Moreover, the parameters of different devices vary greatly, and the image sequence values ​​given by the scanning equipment of different manufacturers are not standardized. The pixel values ​​of DWI and ADC are not necessarily within the standard range and may have undergone complex transformations. Even using conventional normalization such as z-normalization alone cannot transform them into a more consistent distribution, which causes great trouble for segmentation.

[0020] If the processed data is not used in the training process of the deep learning network, it may lead to incorrect segmentation results, unclear edge segmentation, or even missed detections. The accuracy of the output results of the deep learning network can only be corrected by adding a large amount of data to train the model. When the amount of data is insufficient and contains various discrepancies, it is equivalent to indirectly polluting the labeled data, which will lead to a decrease in inference accuracy.

[0021] Based on this, embodiments of this application provide a method for determining lesion images. The specific solutions of these embodiments are described below with reference to the accompanying drawings.

[0022] See Figure 1 The flowchart shown is a method for determining lesion images, which specifically includes the following steps:

[0023] Step 101: Based on the diffusion-weighted image of the target object's head, obtain the apparent diffusion coefficient image. The diffusion-weighted image and the apparent diffusion coefficient image have the same pixel matrix.

[0024] In one embodiment of this application, the subsequences corresponding to diffusion weighting intensities equal to a first preset diffusion weighting intensity threshold and a second preset diffusion weighting intensity threshold in the diffusion weighting image can be separated to obtain multiple first subsequences and multiple second subsequences. Then, each first subsequence and each second subsequence are registered to obtain the apparent diffusion coefficient image.

[0025] It should be noted that diffusion-weighted intensity (b-value) is a parameter used in diffusion-weighted imaging (DWI) to measure diffusion sensitivity, with units of s / mm. 2 Low b-values ​​are more susceptible to perfusion / pseudo-diffusion, while high b-values ​​emphasize molecular diffusion. Commonly used brain scan values ​​are b=0 and b=1000 s / mm. 2 Therefore, in this embodiment of the application, the first preset diffusion weighting intensity threshold is 1000, and the second preset diffusion weighting intensity threshold is 0.

[0026] In the apparent diffusion coefficient image obtained after registering each first subsequence and each second subsequence, each first subsequence and each second subsequence corresponds one-to-one with each pixel of the apparent diffusion coefficient image.

[0027] The above-mentioned separation and registration of the first and second subsequences in the diffusion weighted graph is a commonly used technique in the field, and its implementation process will not be described in detail in this application.

[0028] Based on the above implementation, in some modified implementations, the apparent diffusion coefficient value corresponding to each pixel in the apparent diffusion coefficient image can be calculated using the following formula, based on the signal strength of each first subsequence and the signal strength of the second subsequence:

[0029]

[0030] in, This represents the apparent diffusion coefficient value corresponding to the l-th pixel in the apparent diffusion coefficient image. This represents the signal strength of the l-th first subsequence. b1 represents the signal strength of the l-th second subsequence, b2 represents the first preset diffusion weighting threshold, and b2 represents the second preset diffusion weighting threshold.

[0031] Based on the above implementation, in some modified implementations, the diffusion-weighted image and the apparent diffusion coefficient image are input into a pre-trained lesion segmentation network to obtain the lesion prediction probability value corresponding to each pixel in the diffusion-weighted image or the apparent diffusion coefficient image.

[0032] Step 102: Based on the diffusion-weighted image and the apparent diffusion coefficient image, obtain the lesion-free image and the first candidate lesion image.

[0033] In one embodiment of this application, an adaptive threshold algorithm is used to perform adaptive threshold segmentation on multiple first subsequences in the diffusion-weighted image to obtain a first candidate lesion image. Based on the diffusion-weighted image and the apparent diffusion coefficient image, a brain tissue image is extracted using a brain segmentation algorithm. Then, the first candidate lesion image in the brain tissue image is removed, and morphological operations are performed to obtain a lesion-free image.

[0034] It should be noted that the first candidate lesion image includes multiple sub-candidate lesion images.

[0035] Step 103: Based on the apparent diffusion coefficient value and lesion prediction probability value corresponding to each pixel point of the lesion-free image and the first candidate lesion image, obtain the second candidate lesion image from the first candidate lesion image.

[0036] In one embodiment of this application, the predicted probability value of a lesion corresponding to a first target pixel located at a first preset percentile in a first sub-candidate lesion image and the apparent diffusion coefficient values ​​corresponding to multiple second target pixels located at a second preset percentile are obtained. The first sub-candidate lesion image is any one of the multiple sub-candidate lesion images. The predicted probability value of a lesion corresponding to a third target pixel located at a first preset percentile in a lesion-free image is also obtained. Then, a first ratio is calculated for the percentage of apparent diffusion coefficient values ​​corresponding to each second target pixel in the first sub-candidate lesion image that are greater than or equal to the apparent diffusion coefficient values ​​corresponding to each pixel in the lesion-free image. A second ratio is calculated for the predicted probability value of a lesion corresponding to the first target pixel and the predicted probability value of a lesion corresponding to the third target pixel. If each first ratio satisfies a first preset condition and the second ratio satisfies a second preset condition, then the first sub-candidate lesion image is used as the second candidate lesion image.

[0037] It should be noted that both the first preset percentile and the second preset percentile can be preset by those skilled in the art according to actual needs, or can be obtained by those skilled in the art by adjusting the preset first preset percentile and the second preset percentile respectively according to actual needs. This application embodiment does not impose specific limitations. For example, in this application embodiment, the first preset percentile can be 90, and the second preset percentile can be calculated based on a preset sampling quantity. The specific calculation formula is as follows: ,in, The second preset percentile is represented by i, which represents the preset number of samples. For example, if the preset number of samples is 17, then 17 pixels are obtained from all pixels in the first sub-candidate lesion image. The positions of each obtained pixel are the pixels corresponding to the 10th percentile, 15th percentile, 20th percentile, ..., 90th percentile.

[0038] Furthermore, if the first preset percentile is 90, then the predicted probability value of the lesion corresponding to the first target pixel located at the first preset percentile in the first sub-candidate lesion image can be obtained as follows: First, the predicted probability values ​​of the lesions corresponding to all pixels in the first sub-candidate lesion image are sorted in ascending order. Then, the predicted probability value of the lesion corresponding to the pixel at the 90th position of the first preset percentile is obtained. The 90th position of the first preset percentile is equal to the total number of pixels in the first sub-candidate lesion image multiplied by 90%. That is, if the total number of pixels in the first sub-candidate lesion image is 200, then the pixel at the 90th position of the first percentile is 200 × 90% = 180. In other words, the predicted probability value of the lesion of the 180th pixel in the first sub-candidate lesion image is obtained.

[0039] Similarly, to obtain the apparent diffusion coefficient values ​​corresponding to multiple second target pixels located at the second preset percentile in the first sub-candidate lesion image, the following steps can be taken: first, sort the apparent diffusion coefficient values ​​corresponding to all pixels in the first sub-candidate lesion image in ascending order, and then obtain the apparent diffusion coefficient values ​​corresponding to the pixels at each second preset percentile position. The specific calculation method for the second preset percentile position is the same as that for the first preset percentile position, and will not be repeated here.

[0040] Similarly, to obtain the lesion prediction probability value corresponding to the third target pixel located at the first preset percentile in the lesion-free image, it can be done by first sorting the lesion prediction probability values ​​corresponding to all pixels in the lesion-free image in ascending order, and then obtaining the lesion prediction probability value corresponding to the pixel at the first preset percentile position.

[0041] Further, the first ratio is calculated for each second target pixel in the first sub-candidate lesion image whose apparent diffusion coefficient value is greater than or equal to the apparent diffusion coefficient value of each pixel in the lesion-free image. Specifically, for any apparent diffusion coefficient value corresponding to a second target pixel, it is compared with the apparent diffusion coefficient values ​​of all pixels in the lesion-free image. The total number of pixels in the lesion-free image whose apparent diffusion coefficient value is greater than or equal to the apparent diffusion coefficient value of the second target pixel is recorded as M1, and the total number of pixels in the lesion-free image is recorded as M2. The first ratio is then M1 / M2. Finally, each first ratio can be recorded as... , where i represents the total number of the second target pixels.

[0042] Further, the second ratio of the lesion prediction probability value corresponding to the first target pixel to the lesion prediction probability value corresponding to the third target pixel can be calculated. Specifically, the lesion prediction probability value corresponding to the first target pixel is N1, and the lesion prediction probability value corresponding to the third target pixel is N2. The second ratio is N1 / N2, and the second ratio is denoted as K.

[0043] Based on the above implementation, in some modified implementations, the first preset condition may be: the first ratio is less than or equal to the second preset percentile / 100. If the second preset percentile is denoted as T, then all first ratios satisfying the first preset condition can specifically be: determining whether the i-th first ratio is less than or equal to the i-th second preset percentile / 100, that is, determining... Is it less than or equal to? / 100, if ≤ If the ratio is 100, then it is determined that the i-th first ratio satisfies the first preset condition.

[0044] Based on the above implementation method, in some modified implementation methods, the second preset condition can be: the second logarithmic ratio to the base 10 is greater than a preset threshold. Specifically, it is determined whether the second logarithmic ratio to the base 10 is greater than the preset threshold. If the second logarithmic ratio to the base 10 is greater than the preset threshold, it is determined that the second ratio satisfies the second preset condition.

[0045] It should be noted that the preset threshold can be a value pre-set by those skilled in the art according to actual needs, or a value obtained by those skilled in the art after adjusting a pre-set value according to actual needs. This application embodiment does not impose specific limitations. In this application embodiment, the preset threshold can be a value greater than zero, for example, the preset threshold can be 1.669.

[0046] By combining a deep learning network model (lesion segmentation network model) to calculate the lesion prediction probability value of the image, and then based on the apparent diffusion coefficient value and lesion prediction probability value corresponding to each pixel of the lesion-free image and the first candidate lesion image, a second candidate lesion image is obtained from the first candidate lesion image. This method does not rely entirely on the results of the deep learning network model, thus enabling compatibility with data from more device manufacturers, including data from mainstream device manufacturers and non-mainstream manufacturers, effectively improving the compatibility of the lesion image determination process.

[0047] Step 104: If the rate of change between the second candidate lesion image and the third candidate lesion image obtained in the previous iteration exceeds the preset rate of change, then proceed to step 105; otherwise, proceed to step 106.

[0048] In one embodiment of this application, the rate of change may include whether the second candidate lesion image is the same as the third candidate lesion image obtained in the previous iteration. If the second candidate lesion image is the same as the third candidate lesion image obtained in the previous iteration, it is determined that the rate of change between the second candidate lesion image and the third candidate lesion image obtained in the previous iteration does not exceed a preset rate of change, and the second candidate lesion image is determined to be a lesion image.

[0049] Based on the above implementation methods, in some modified implementation methods, the rate of change may further include the rate of change between the number of pixels in the second candidate lesion image and the number of pixels in the third candidate lesion image obtained in the previous iteration. That is, first, the rate of change between the number of pixels in the second candidate lesion image and the number of pixels in the third candidate lesion image obtained in the previous iteration is obtained. This rate of change is the ratio of the difference between the number of pixels in the second candidate lesion image and the number of pixels in the third candidate lesion image obtained in the previous iteration to the number of pixels in the second candidate lesion image. Then, it is determined whether this rate of change is less than a preset rate of change. If it is less, it is determined that the rate of change between the second candidate lesion image and the third candidate lesion image obtained in the previous iteration does not exceed the preset rate of change, and the second candidate lesion image is then identified as an image with a lesion. It should be noted that the preset rate of change in this embodiment can be a rate of change value preset by those skilled in the art according to actual needs, or it can be obtained by those skilled in the art after adjusting a preset rate of change value according to actual needs. This embodiment does not impose specific limitations. For example, the rate of change value in this embodiment can be 2%.

[0050] Step 105: Expand the first candidate lesion image to its current preset number of times, and continue to step 103.

[0051] In one embodiment of this application, the first candidate lesion image includes multiple sub-candidate lesion images. Specifically, expanding the first candidate lesion image to its current preset number of times can be achieved by: obtaining the smallest rectangular bounding box that can contain each sub-candidate lesion image, and then expanding the length and width to the original preset number of times with the centroid of the rectangular bounding box as the center.

[0052] It should be noted that the preset number refers to the magnification factor of each sub-candidate lesion image. The preset number can be a factor preset by those skilled in the art according to actual needs, or it can be a factor obtained by adjusting a preset factor according to actual needs. This application embodiment does not impose a specific limitation. For example, the factor in this application embodiment can be 2.

[0053] Step 106: Determine the second candidate lesion image as an image with lesions.

[0054] The lesion image determination method of this application embodiment includes the following steps: Step 1: Obtaining an apparent diffusion coefficient image based on a diffusion-weighted image of the target object's head, wherein the diffusion-weighted image and the apparent diffusion coefficient image have the same pixel matrix; Step 2: Obtaining a lesion-free image and a first candidate lesion image based on the diffusion-weighted image and the apparent diffusion coefficient image; Step 3: Obtaining a second candidate lesion image from the first candidate lesion image based on the apparent diffusion coefficient value and lesion prediction probability value corresponding to each pixel point of the lesion-free image and the first candidate lesion image; Step 4: If the rate of change between the second candidate lesion image and the third candidate lesion image obtained in the previous iteration exceeds a preset rate of change, then proceed to Step 5; otherwise, the second candidate lesion image is determined to be a lesion image; Step 5: Expanding the first candidate lesion image to its current preset number of times and continuing to execute Step 3, which can improve the success rate of lesion determination under poor data compatibility, continuously correct the determination result of the candidate lesion image by iterative method, avoid the result error caused by using a single threshold, and improve the accuracy of lesion determination.

[0055] To provide a coherent and clear explanation of the lesion image determination method of this application's embodiments, the following overall description uses examples:

[0056] 1. Obtain the diffusion-weighted image (DWI image) of the target object's head, separate the subsequence B1000 (diffusion weighting intensity b equals the first preset diffusion weighting intensity threshold of 1000) and the subsequence B0 (diffusion weighting intensity b equals the second preset diffusion weighting intensity threshold of 0) in the DWI image, and obtain multiple subsequences B1000 and multiple subsequences B0. Perform registration processing on each subsequence B1000 and each subsequence B0 to obtain the apparent diffusion coefficient image (ADC image). The diffusion-weighted image and the apparent diffusion coefficient image have the same pixel matrix.

[0057] Then, based on the signal strength of each subsequence B1000 and the signal strength of subsequence B0, the apparent diffusion coefficient value corresponding to each pixel in the apparent diffusion coefficient image is calculated using the following formula:

[0058]

[0059] in, This represents the apparent diffusion coefficient value corresponding to the l-th pixel in the apparent diffusion coefficient image. This represents the signal strength of the l-th subsequence of the first subsequence. b1 represents the signal strength of the l-th second sub-sequence, b2 represents the first preset diffusion weighting threshold, and b2 represents the second preset diffusion weighting threshold.

[0060] 2. Using an adaptive thresholding method, high-signal regions with B1000 in multiple sub-sequences are calculated and denoted as high B1000 regions (first candidate lesion images). Based on DWI and ADC images, brain tissue images are extracted using a brain segmentation algorithm. The high B1000 regions in the brain tissue images are removed, and morphological operations are performed to obtain normal brain tissue images (lesion-free images). Then, using this normal brain tissue image as a mask, the ADC values ​​corresponding to the pixels within the mask in the ADC image are obtained, which are the ADC values ​​corresponding to each pixel in the lesion-free image.

[0061] 3. Input the DWI image and ADC image into the pre-trained lesion segmentation network model to obtain the lesion prediction probability value corresponding to each pixel of the image. Using the normal brain tissue image as a mask, obtain the lesion prediction probability value corresponding to each pixel within the mask, which is the lesion prediction probability value corresponding to each pixel of the lesion-free image.

[0062] 4. Merge multiple sub-candidate lesion images in the first candidate lesion image into a candidate lesion image set.

[0063] 5. For each sub-candidate lesion image in the candidate lesion image set, determine the following two conditions:

[0064] a) Condition 1:

[0065] i. Obtain the ADC value of each pixel in the first sub-candidate lesion image;

[0066] ii. Obtain the i-th element from the first sub-candidate lesion image respectively. The ADC value corresponding to the second target pixel at the percentile is denoted as... ,in , i represents the number of samples;

[0067] iii. by each Using the threshold value, the ADC value of each pixel in the lesion-free image is calculated to be less than... The number of pixels, denoted as The total number of pixels in the lesion-free image is denoted as M2, and then the ratios are obtained: ;

[0068] iv. Compare separately and / 100, if all All less than or equal to If the value is 100, then it is considered to satisfy condition 1.

[0069] b) Condition 2:

[0070] i. Obtain the predicted probability value of the lesion corresponding to the first target pixel point corresponding to the Lth percentile in the first sub-candidate lesion image, denoted as N1;

[0071] ii. Obtain the lesion prediction probability value corresponding to the third target pixel point at the Lth percentile in the lesion-free image, denoted as N2, where In this example, L=90;

[0072] iii. If In this example, t=1.669, which is considered to satisfy condition 2.

[0073] Traverse all candidate lesion images and retain only those that satisfy both condition 1 and condition 2 as a new set of candidate lesion images. This new set of candidate lesion images constitutes the second candidate lesion image.

[0074] 6. Enlarge the first candidate lesion image to twice its current size, that is, obtain the smallest rectangular bounding box that can contain each sub-candidate lesion image, and then enlarge the length and width to twice their current size with the centroid of the bounding box as the center.

[0075] 7. Repeat steps 4, 5, and 6 until the second candidate lesion image obtained in step 5 no longer changes, or the rate of change of the number of pixels in the second candidate lesion image is less than 2%. The second candidate lesion image at the time when the iteration stops is determined to be a lesion image.

[0076] See Figure 2 This application also provides a lesion image determination device, which is used to perform the lesion image determination method described in the above embodiments. The device includes:

[0077] The first acquisition module 201 is used to acquire an apparent diffusion coefficient image based on the diffusion-weighted image of the head of the target object, wherein the diffusion-weighted image and the apparent diffusion coefficient image have the same pixel matrix.

[0078] The second acquisition module 202 is used to acquire a lesion-free image and a first candidate lesion image based on the diffusion-weighted image and the apparent diffusion coefficient image;

[0079] The third acquisition module 203 is used to acquire a second candidate lesion image from the first candidate lesion image based on the apparent diffusion coefficient value and lesion prediction probability value corresponding to each pixel point of the lesion-free image and the first candidate lesion image.

[0080] The lesion determination module 204 is used to jump to the image processing module 205 if the rate of change between the second candidate lesion image and the third candidate lesion image obtained from the previous iteration exceeds a preset rate of change; otherwise, the second candidate lesion image is determined to be an image with lesions.

[0081] Image processing module 205 is used to enlarge the first candidate lesion image to a preset number of times and then jump to the third acquisition module 203.

[0082] The lesion image determination device provided in this application embodiment is based on the same inventive concept as the lesion image determination method provided in the above embodiment, and has the same beneficial effects as the method used, operated or implemented therein.

[0083] This application also provides an electronic device corresponding to the lesion image determination method provided in the foregoing embodiments. Please refer to... Figure 3 This illustrates a schematic diagram of an electronic device provided by some embodiments of this application. For example... Figure 3 As shown, the electronic device 30 may include: a processor 300, a memory 301, a bus 302, and a communication interface 303. The processor 300, the communication interface 303, and the memory 301 are connected via the bus 302. The memory 301 stores a computer program that can run on the processor 300. When the processor 300 runs the computer program, it executes the lesion image determination method provided in any of the foregoing embodiments of this application.

[0084] The memory 301 may include high-speed random access memory (RAM) or non-volatile memory, such as at least one disk storage device. Communication between this system network element and at least one other network element is achieved through at least one communication port 303 (which can be wired or wireless), such as the Internet, wide area network, local area network, or metropolitan area network.

[0085] Bus 302 can be an ISA bus, PCI bus, or EISA bus, etc. The bus can be divided into an address bus, a data bus, a control bus, etc. The memory 301 is used to store programs. After receiving an execution instruction, the processor 300 executes the program. The lesion image determination method disclosed in any of the foregoing embodiments of this application can be applied to the processor 300, or implemented by the processor 300.

[0086] The processor 300 may be an integrated circuit with signal processing capabilities. In implementation, each step of the above method can be completed by the integrated logic circuitry in the hardware of the processor 300 or by instructions in software form. The processor 300 may be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it may also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), an off-the-shelf programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. It can implement or execute the methods, steps, and logic block diagrams disclosed in the embodiments of this application. The general-purpose processor may be a microprocessor or any conventional processor. The steps of the methods disclosed in the embodiments of this application can be directly embodied in the execution of a hardware decoding processor, or executed by a combination of hardware and software modules in the decoding processor. The software modules may reside in random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, registers, or other mature storage media in the art. The storage medium is located in memory 301. The processor 300 reads the information in memory 301 and, in conjunction with its hardware, completes the steps of the above method.

[0087] The electronic device provided in this application embodiment and the lesion image determination method provided in this application embodiment are based on the same inventive concept and have the same beneficial effects as the methods they adopt, operate or implement.

[0088] This application also provides a computer-readable storage medium corresponding to the lesion image determination method provided in the foregoing embodiments, wherein a computer program (i.e., a program product) is stored thereon, and the computer program, when run by a processor, executes the lesion image determination method provided in any of the foregoing embodiments.

[0089] It should be noted that examples of the computer-readable storage medium may also include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other optical and magnetic storage media, which will not be elaborated here.

[0090] This application also provides a computer program product corresponding to the lesion image determination method provided in the foregoing embodiments, including a computer program that is executed by a processor to implement the lesion image determination method provided in the above embodiments.

[0091] The computer-readable storage medium and computer program product provided in the above embodiments of this application are based on the same inventive concept as the lesion image determination method provided in the embodiments of this application, and have the same beneficial effects as the methods adopted, run or implemented by the application programs stored therein.

[0092] It should be noted that:

[0093] The algorithms and displays provided herein are not inherently related to any particular computer, virtual device, or other equipment. Various general-purpose devices can also be used in conjunction with the teachings herein. The required structure for constructing such devices is apparent from the above description. Furthermore, this application is not directed to any particular programming language. It should be understood that the content of this application described herein can be implemented using various programming languages, and the above description of specific languages ​​is for the purpose of disclosing the best mode of implementation of this application.

[0094] Numerous specific details are set forth in the specification provided herein. However, it will be understood that embodiments of this application may be practiced without these specific details. In some instances, well-known methods, structures, and techniques have not been shown in detail so as not to obscure the understanding of this specification.

[0095] Similarly, it should be understood that, for the purpose of simplification and aiding understanding of one or more aspects of the invention, various features of the present application are sometimes grouped together in a single embodiment, figure, or description thereof in the description of exemplary embodiments of the present application above. Those skilled in the art will understand that modules in the apparatus of an embodiment can be adaptively changed and disposed in one or more devices different from that embodiment. Modules, units, or components in an embodiment can be combined into a single module, unit, or component, and further, they can be divided into multiple sub-modules, sub-units, or sub-components. Except where at least some of such features and / or processes or units are mutually exclusive, any combination can be employed to combine all features disclosed in this specification and all processes or units of any method or apparatus so disclosed. Unless expressly stated otherwise, each feature disclosed in this specification may be replaced by an alternative feature that serves the same, equivalent, or similar purpose.

[0096] Furthermore, those skilled in the art will understand that although some embodiments described herein include certain features included in other embodiments but not others, combinations of features from different embodiments are intended to be within the scope of this application and form different embodiments. The various component embodiments of this application can be implemented in hardware, or as software modules running on one or more processors, or a combination thereof. Those skilled in the art will understand that microprocessors or digital signal processors (DSPs) can be used in practice to implement some or all of the functions of some or all of the components in the virtual machine creation apparatus according to embodiments of this application. This application can also be implemented as a device or apparatus program (e.g., a computer program and computer program product) for performing part or all of the methods described herein. Such an implementation of this application can be stored on a computer-readable medium or can be in the form of one or more signals. Such signals can be downloaded from an Internet website, provided on a carrier signal, or provided in any other form.

[0097] The above description is merely a preferred embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application.

Claims

1. A method for determining lesion images, characterized in that, include: Step 1: Based on the diffusion-weighted image of the target object's head, obtain the apparent diffusion coefficient image, wherein the diffusion-weighted image and the apparent diffusion coefficient image have the same pixel matrix; Step 2: Based on the diffusion-weighted image and the apparent diffusion coefficient image, obtain the lesion-free image and the first candidate lesion image; Step 3: Based on the apparent diffusion coefficient value and lesion prediction probability value corresponding to each pixel point of the lesion-free image and the first candidate lesion image, obtain the second candidate lesion image from the first candidate lesion image; Step 4: Enlarge the first candidate lesion image to a preset number of times. Based on the apparent diffusion coefficient value and lesion prediction probability value corresponding to each pixel of the lesion-free image and the enlarged first candidate lesion image, obtain the second candidate lesion image from the first candidate lesion image. Step 5: If the rate of change between the second candidate lesion image in the current round and the second candidate lesion image calculated in the previous iteration exceeds the preset rate of change, then proceed to step 4; otherwise, determine the second candidate lesion image in the current round as an image with lesions. The first candidate lesion image includes multiple sub-candidate lesion images. The step of obtaining a second candidate lesion image from the first candidate lesion image based on the apparent diffusion coefficient value and lesion prediction probability value corresponding to each pixel point of the lesion-free image and the first candidate lesion image includes: Obtain the predicted probability value of the lesion corresponding to the first target pixel located at the first preset percentile in the first sub-candidate lesion image and the apparent diffusion coefficient values ​​corresponding to multiple second target pixels located at the second preset percentile, wherein the first sub-candidate lesion image is any one of the multiple sub-candidate lesion images; Obtain the predicted probability value of the lesion corresponding to the third target pixel located at the first preset percentile in the lesion-free image; Calculate the first ratio of the number of pixels in the first sub-candidate lesion image whose apparent diffusion coefficient value is greater than or equal to the apparent diffusion coefficient value of each pixel in the lesion-free image. Calculate a second ratio between the predicted probability value of the lesion corresponding to the first target pixel and the predicted probability value of the lesion corresponding to the third target pixel; If each of the first ratios satisfies the first preset condition, and the second ratio satisfies the second preset condition, then the first sub-candidate lesion image is used as the second candidate lesion image.

2. The lesion image determination method according to claim 1, characterized in that, The first preset condition is: the first ratio is less than or equal to the second preset percentile / 100.

3. The lesion image determination method according to claim 1, characterized in that, The second preset condition is: the second ratio, which is the base-10 logarithm, is greater than a preset threshold.

4. The method for determining lesion images according to claim 1, characterized in that, The apparent diffusion coefficient image includes multiple first sub-sequences and multiple second sub-sequences. The diffusion weighting intensity of each first sub-sequence is equal to a first preset diffusion weighting intensity threshold, and the diffusion weighting intensity of each second sub-sequence is equal to a second preset diffusion weighting intensity threshold. Before obtaining the lesion-free image and the first candidate lesion image based on the diffusion weighting image and the apparent diffusion coefficient image, the method further includes: Based on the signal strength of each of the first subsequences and the signal strength of the second subsequences, the apparent diffusion coefficient value corresponding to each pixel in the apparent diffusion coefficient image is calculated using the following formula: in, This represents the apparent diffusion coefficient value corresponding to the l-th pixel in the apparent diffusion coefficient image. This represents the signal strength of the l-th subsequence of the first subsequence. b1 represents the signal strength of the l-th second sub-sequence, b2 represents the first preset diffusion weighting threshold, and b2 represents the second preset diffusion weighting threshold.

5. The method for determining lesion images according to claim 1, characterized in that, Before obtaining the lesion-free image and the first candidate lesion image based on the diffusion-weighted image and the apparent diffusion coefficient image, the method further includes: The diffusion-weighted image and the apparent diffusion coefficient image are input into a pre-trained lesion segmentation network model to obtain the lesion prediction probability value corresponding to each pixel in the diffusion-weighted image or the apparent diffusion coefficient image.

6. The method for determining lesion images according to claim 1, characterized in that, The rate of change includes whether the second candidate lesion image in the current round is the same as the second candidate lesion image calculated in the previous iteration, and / or the rate of change of the number of pixels in the second candidate lesion image in the current round compared with the number of pixels in the second candidate lesion image calculated in the previous iteration.

7. A lesion image determination device, characterized in that, include: The first acquisition module is used to acquire an apparent diffusion coefficient image based on the diffusion-weighted image of the target object's head, wherein the diffusion-weighted image and the apparent diffusion coefficient image have the same pixel matrix. The second acquisition module is used to acquire a lesion-free image and a first candidate lesion image based on the diffusion-weighted image and the apparent diffusion coefficient image. The third acquisition module is used to acquire a second candidate lesion image from the first candidate lesion image based on the apparent diffusion coefficient value and lesion prediction probability value corresponding to each pixel point of the lesion-free image and the first candidate lesion image. The first candidate lesion image includes multiple sub-candidate lesion images. The image processing module is used to enlarge the first candidate lesion image to a preset number of times its current size, and to obtain a second candidate lesion image from the first candidate lesion image based on the apparent diffusion coefficient value and lesion prediction probability value corresponding to each pixel point of the lesion-free image and the enlarged first candidate lesion image. The lesion determination module is used to jump to the image processing module if the rate of change between the second candidate lesion image in the current round and the second candidate lesion image calculated in the previous iteration exceeds a preset rate of change; otherwise, the second candidate lesion image in the current round is determined to be an image with lesions. The third acquisition module is used to acquire the lesion prediction probability value corresponding to the first target pixel point located at the first preset percentile in the first sub-candidate lesion image and the apparent diffusion coefficient value corresponding to multiple second target pixels located at the second preset percentile, wherein the first sub-candidate lesion image is any one of the multiple sub-candidate lesion images; and to acquire the lesion prediction probability value corresponding to the third target pixel point located at the first preset percentile in the lesion-free image. Calculate the first ratio of the number of pixels in the first sub-candidate lesion image whose apparent diffusion coefficient value is greater than or equal to the apparent diffusion coefficient value of each pixel in the lesion-free image. Calculate a second ratio between the predicted probability value of the lesion corresponding to the first target pixel and the predicted probability value of the lesion corresponding to the third target pixel; If each of the first ratios satisfies the first preset condition, and the second ratio satisfies the second preset condition, then the first sub-candidate lesion image is used as the second candidate lesion image.

8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the method according to any one of claims 1-6.

9. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by a processor, it implements the method described in any one of claims 1-6.

Images