Medical information analysis method and device suitable for medical image

Abstract

The invention relates to a medical information analysis method and device suitable for a medical image. The method comprises the following steps: acquiring a multi-contrast image of a target part; the multi-contrast images comprise a water distribution image, a fat distribution image, an in-phase image and an anti-phase image; inputting the water distribution image, the fat distribution image, the in-phase image and the anti-phase image into a trained target recognition model to obtain a recognition result of a set tissue of the target part; and displaying the quantitative analysis result of the set tissue of the target part in a visual form according to the recognition result of the set tissue. By adopting the method, the accuracy of medical information analysis can be improved.

Description

Technical Field

[0001] The present application relates to the technical field of medical image processing, and in particular, to a medical information analysis method and device applicable to medical images. Background Art

[0002] Medical information analysis is of great significance in clinical treatment. For example, during the rehabilitation training process, it is usually necessary to understand the muscle growth or changes in a specific part of the patient in order to adjust the rehabilitation training plan according to the muscle growth or changes.

[0003] Currently, visual observation is usually used to subjectively evaluate the muscle growth or changes. This method is difficult to quantitatively determine which muscles have grown due to exercise and which muscles have been compensated due to incorrect postures and have experienced abnormal growth. Furthermore, it is impossible to accurately correct the exercise methods and movements.

[0004] Therefore, there is a problem of low accuracy in the current medical information analysis technology. Summary of the Invention

[0005] Based on this, in view of the above technical problems, it is necessary to provide a relatively accurate medical information analysis method, device, computer device, computer-readable storage medium, and computer program product.

[0006] In a first aspect, the present application provides a medical information analysis method applicable to medical images. The method includes:

[0007] Obtain multi-contrast images of a target part; the multi-contrast images include a water distribution image, a fat distribution image, an in-phase image, and an opposed-phase image;

[0008] Input the water distribution image, the fat distribution image, the in-phase image, and the opposed-phase image into a trained target recognition model to obtain the recognition result of a set tissue of the target part;

[0009] According to the recognition result of the set tissue, display the quantitative analysis result of the set tissue of the target part in a visual form.

[0010] In one embodiment, before obtaining the multi-contrast images of the target part, it further includes:

[0011] Obtain multi-contrast image samples of the target part; the multi-contrast image samples include a water distribution image sample, a fat distribution image sample, an in-phase image sample, and an opposed-phase image sample;

[0012] Train the target recognition model to be trained according to the multi-contrast image samples and the image sample labels corresponding to the multi-contrast image samples, to obtain the trained target recognition model.

[0013] In one embodiment, after obtaining the multi-contrast image samples of the target part, it further includes:

[0014] Segment the set tissue in the water distribution image sample, the fat distribution image sample, the in-phase image sample and the opposed-phase image sample respectively, to obtain the first sample segmentation result, the second sample segmentation result, the third sample segmentation result and the fourth sample segmentation result of the set tissue;

[0015] Fuse the first sample segmentation result, the second sample segmentation result, the third sample segmentation result and the fourth sample segmentation result to obtain a fused sample segmentation result;

[0016] Obtain the image sample label according to the fused sample segmentation result.

[0017] In one embodiment, the fusing the first sample segmentation result, the second sample segmentation result, the third sample segmentation result and the fourth sample segmentation result to obtain a fused sample segmentation result includes:

[0018] Determine an initial sample segmentation result from the first sample segmentation result, the second sample segmentation result, the third sample segmentation result and the fourth sample segmentation result, and determine the sample segmentation results other than the initial sample segmentation result as adjusted sample segmentation results;

[0019] Adjust the initial sample segmentation result according to each of the adjusted sample segmentation results respectively, to obtain the fused sample segmentation result.

[0020] In one embodiment, the visualizing the quantitative analysis result of the set tissue of the target part according to the recognition result of the set tissue includes:

[0021] Obtain the change curve of the tissue parameters of the set tissue according to the recognition results of the set tissue corresponding to at least one target time;

[0022] Display the change curve of the tissue parameters.

[0023] In one embodiment, the tissue parameters include volume; the obtaining the change curve of the tissue parameters of the set tissue according to the recognition results of the set tissue corresponding to at least one target time includes:

[0024] Determine the volume of the set tissue at each of the target times according to the recognition result of the set tissue;

[0025] Obtain the volume change curve of the set tissue according to the volume at each of the target times.

[0026] In one embodiment, the tissue parameter further includes a volume ratio; the obtaining the change curve of the tissue parameter of the set tissue according to the recognition result of the set tissue corresponding to at least one target time further includes:

[0027] Determine the volume ratio of the set tissue at the target site at each of the target times according to the recognition result of the set tissue;

[0028] Obtain the volume ratio change curve of the set tissue according to the volume ratio at each of the target times.

[0029] In a second aspect, the present application further provides a medical information analysis device applicable to medical images. The device includes:

[0030] An acquisition module, configured to acquire multi-contrast images of a target site; the multi-contrast images include a water distribution image, a fat distribution image, an in-phase image, and an opposed-phase image;

[0031] A recognition module, configured to input the water distribution image, the fat distribution image, the in-phase image, and the opposed-phase image into a trained target recognition model to obtain a recognition result of a set tissue of the target site;

[0032] A display module, configured to visually display a quantitative analysis result of the set tissue of the target site according to the recognition result of the set tissue.

[0033] In a third aspect, the present application further provides a computer device. The computer device includes a memory and a processor, the memory stores a computer program, and when the processor executes the computer program, the following steps are implemented:

[0034] Acquire multi-contrast images of a target site; the multi-contrast images include a water distribution image, a fat distribution image, an in-phase image, and an opposed-phase image;

[0035] Input the water distribution image, the fat distribution image, the in-phase image, and the opposed-phase image into a trained target recognition model to obtain a recognition result of a set tissue of the target site;

[0036] Visually display a quantitative analysis result of the set tissue of the target site according to the recognition result of the set tissue.

[0037] Fourthly, the present application also provides a computer-readable storage medium. On the computer-readable storage medium, there is a computer program which, when executed by a processor, implements the following steps:

[0038] Obtain multi-contrast images of a target part; the multi-contrast images include a water distribution image, a fat distribution image, an in-phase image, and an opposed-phase image;

[0039] Input the water distribution image, the fat distribution image, the in-phase image, and the opposed-phase image into a trained target recognition model to obtain a recognition result of a set tissue of the target part;

[0040] According to the recognition result of the set tissue, visually display a quantitative analysis result of the set tissue of the target part.

[0041] Fifthly, the present application also provides a computer program product. The computer program product includes a computer program which, when executed by a processor, implements the following steps:

[0042] Obtain multi-contrast images of a target part; the multi-contrast images include a water distribution image, a fat distribution image, an in-phase image, and an opposed-phase image;

[0043] Input the water distribution image, the fat distribution image, the in-phase image, and the opposed-phase image into a trained target recognition model to obtain a recognition result of a set tissue of the target part;

[0044] According to the recognition result of the set tissue, visually display a quantitative analysis result of the set tissue of the target part.

[0045] The above medical information analysis method, device, computer device, storage medium, and computer program product applicable to medical images obtain multi-contrast images of a target part, where the multi-contrast images include a water distribution image, a fat distribution image, an in-phase image, and an opposed-phase image, input the water distribution image, the fat distribution image, the in-phase image, and the opposed-phase image into a trained target recognition model to obtain a recognition result of a set tissue of the target part, and according to the recognition result of the set tissue, visually display a quantitative analysis result of the set tissue of the target part; it can identify set tissues such as total body fat, muscle, bone, liver, uterus, prostate, etc. based on the multi-contrast images, improve the accuracy of the recognition result of the set tissue, and thus improve the accuracy of medical information analysis. Visually displaying the quantitative analysis result of the set tissue also increases the readability of the medical information analysis result. Description of the Drawings

[0046] Figure 1Schematic flowchart of a medical information analysis method applicable to medical images in an embodiment;

[0047] Figure 2 Schematic diagram of a multi-contrast image in an embodiment;

[0048] Figure 3 Schematic diagram of a multi-contrast image of the shoulder and upper arm in an embodiment;

[0049] Figure 4 Schematic diagram of a multi-contrast image of the shoulder and upper arm in another embodiment;

[0050] Figure 5 Schematic diagram of muscle recognition results in an embodiment;

[0051] Figure 6 Schematic flowchart of the training process of a target recognition model in an embodiment;

[0052] Figure 7 Schematic flowchart of the recognition process of a target recognition model in an embodiment;

[0053] Figure 8 Schematic diagram of a muscle volume change curve in an embodiment;

[0054] Figure 9 Schematic diagram of a three-dimensional scan imaging result in an embodiment;

[0055] Figure 10 Schematic flowchart of a medical information analysis method applicable to medical images in another embodiment;

[0056] Figure 11 Schematic block diagram of a medical information analysis device applicable to medical images in an embodiment;

[0057] Figure 12 Internal structure diagram of a computer device in an embodiment. Detailed implementation manners

[0058] In order to make the objectives, technical solutions and advantages of the present application clearer and more understandable, the present application will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0059] In one embodiment, as Figure 1 shown, a medical information analysis method applicable to medical images is provided. In this embodiment, the method is illustrated by taking its application to a terminal as an example. It can be understood that the method can also be applied to a server, and can also be applied to a system including a terminal and a server, and is implemented through the interaction between the terminal and the server. In this embodiment, the method includes the following steps:

[0060] Step S110, obtain multi-contrast images of the target site; the multi-contrast images include a water distribution image, a fat distribution image, an in-phase image, and an opposed-phase image.

[0061] Among them, the target site can be the site targeted for medical information analysis. For example, it can be the shoulder or upper limb site where muscle changes need to be analyzed.

[0062] Among them, the multi-contrast images can be multiple medical images with different contrasts. The water distribution image is a pure water image, which can be obtained by using a water excitation method or a fat saturation method. The fat distribution image is a pure fat image. In practical applications, a Dixon imaging method can be used to simultaneously obtain an in-phase image and two opposed-phase images. Subsequently, the additional phase caused by the inhomogeneity of the external magnetic field is obtained from the two opposed-phase images, and the two opposed-phase images are phase-corrected. Then, the water distribution image and the fat distribution image are obtained from the in-phase image respectively. It is also possible to use a water-fat separation method based on the inversion recovery (IR) technique: use a radiofrequency pulse to excite the target site to simultaneously excite the hydrogen nuclei in water and fat; determine the inversion recovery time TI of fat, and this TI time is set so that the hydrogen nuclei in fat only have a transverse magnetization component; at the TI time, use an imaging sequence to excite the target site, and only the hydrogen atoms in water will be excited again. The magnetic resonance signals collected are reconstructed to obtain the water distribution image. Based on a similar method, the fat distribution image can also be obtained.

[0063] Among them, the in-phase image can be a magnetic resonance image obtained when the transverse magnetization vectors of water and fat coincide, and the opposed-phase image can be a magnetic resonance image obtained when the transverse magnetization vectors of water and fat are in opposite directions. Specifically, the in-phase image means that when a phase encoding gradient is applied during scanning, the water and fat signals have the same phase, so they show the same brightness in the image. The in-phase image can be used to display the morphology and position information of tissues, but it cannot separate the water and fat signals. The opposed-phase image is when an opposed-phase encoding gradient is applied during scanning, and the water and fat signals have opposite phases, so they show different brightnesses in the image. The in-phase image and the opposed-phase image have different functions in the MRI water-fat separation technology, and their combined use can provide more comprehensive and accurate tissue composition information.

[0064] In specific implementation, a medical imaging device can be used to collect multi-contrast images of the target site, including a water distribution image, a fat distribution image, an in-phase image, and an opposed-phase image. The medical imaging device inputs the collected multi-contrast images to the terminal, so that the terminal obtains the multi-contrast images of the target site.

[0065] Figure 2 A schematic diagram of a multi-contrast image is provided. According to Figure 2 , a water-fat separation technique can be adopted. Based on the characteristics that fat and water are at different resonance frequencies, combined with a three-dimensional gradient echo sequence, an image can be obtained by a single scan, such as Figure 2 The water distribution image (upper left), fat distribution image (upper right), in-phase image (lower left), and opposed-phase image (lower right) shown. Optionally, the in-phase image can be reconstructed using in-phase echo data; the opposed-phase image can be reconstructed using two sets of opposed-phase echo data. During the actual scanning process, the in-phase echo data can be acquired first, and then the two sets of opposed-phase echo data can be acquired; or the in-phase echo data can be obtained during the acquisition interval of the two sets of opposed-phase echo data.

[0066] Step S120: Input the water distribution image, fat distribution image, in-phase image, and opposed-phase image into the trained target recognition model to obtain the recognition result of the set tissue of the target part.

[0067] Among them, the target recognition model can be an AI (Artificial Intelligence) model for image segmentation of multi-contrast images, including but not limited to neural network models and general large models.

[0068] Among them, the set tissue can be the tissue of the preset target part, for example, a combination of one or more of muscle tissue, liver, uterus, and prostate.

[0069] In a specific implementation, the terminal can pre-train the target recognition model, input the water distribution image, fat distribution image, in-phase image, and opposed-phase image into the trained target recognition model to obtain the first segmentation result of the set tissue in the water distribution image, the second segmentation result in the fat distribution image, the third segmentation result in the in-phase image, and the fourth segmentation result in the opposed-phase image, fuse the first segmentation result, the second segmentation result, the third segmentation result, and the fourth segmentation result, and determine the recognition result of the set tissue of the target part according to the obtained fused segmentation result.

[0070] Figure 3 A schematic diagram of a multi-contrast image of the shoulder and upper arm is provided. According to Figure 3 , in the multi-contrast image, the fat distribution image (upper right) and the opposed-phase image (lower right) show the clearest boundaries between muscle and bone.

[0071] Figure 4 Another schematic diagram of a multi-contrast image of the shoulder and upper arm is provided. According to Figure 4, in the multi-contrast images, the opposed-phase image (lower right) shows the clearest boundary between the two muscles, the water distribution image (upper left) shows the clearest lesion contour, and the lesion composition needs to be analyzed by combining the water distribution image with the in-phase image (lower left). Therefore, the muscle segmentation boundary and disease diagnosis results obtained by comprehensively considering the multi-contrast images are the most accurate.

[0072] Figure 5 A schematic diagram of the muscle recognition result is provided. According to Figure 5 , the muscles in the water distribution image, fat distribution image, in-phase image, and opposed-phase image can be segmented respectively to obtain 4 segmentation results of the muscles. The 4 segmentation results are fused to obtain the shoulder muscle recognition result A and upper arm muscle recognition result B as shown in Figure 5 . Different colors or different grayscales can be used to mark the shoulder muscle recognition result A and upper arm muscle recognition result B.

[0073] Step S130, according to the recognition result of the set tissue, display the quantitative analysis result of the set tissue of the target part in a visual form.

[0074] Among them, the quantitative analysis result can be the result obtained by quantitatively analyzing the set tissue. For example, the volume of the set tissue or the proportion of the set tissue in the target part.

[0075] Among them, the visual form can be a graphical display, color-coded display, etc. of the quantitative analysis result.

[0076] In specific implementation, the terminal can quantitatively analyze the obtained recognition result to obtain the quantitative analysis result of the set tissue, and can also display the quantitative analysis result in a visual form to increase the readability of the quantitative analysis result.

[0077] For example, at multiple moments, according to the shoulder muscle recognition result, the volume or volume proportion of the shoulder muscle can be statistically analyzed, a timeline including the above multiple moments can be drawn, and the volume or volume proportion of the shoulder muscle changing with time can be displayed in the form of a curve on the timeline.

[0078] The above medical information analysis method applicable to medical images obtains multi-contrast images of the target part. The multi-contrast images include water distribution images, fat distribution images, in-phase images, and opposed-phase images. The water distribution images, fat distribution images, in-phase images, and opposed-phase images are input into a trained target recognition model to obtain the recognition result of the set tissue of the target part. According to the recognition result of the set tissue, the quantitative analysis result of the set tissue of the target part is displayed in a visual form. It is possible to identify set tissues such as total body fat, muscle, bone, liver, uterus, prostate, etc. based on multi-contrast images, improve the accuracy of the recognition result of the set tissue, and thus improve the accuracy of medical information analysis. Visualizing the quantitative analysis result of the set tissue also increases the readability of the medical information analysis result.

[0079] In one embodiment, before the above step S110, it may specifically further include: obtaining a multi-contrast image sample of the target part; the multi-contrast image sample includes a water distribution image sample, a fat distribution image sample, an in-phase image sample, and an opposed-phase image sample; training the target recognition model to be trained according to the multi-contrast image sample and the image sample label corresponding to the multi-contrast image sample to obtain a trained target recognition model.

[0080] Among them, the multi-contrast image sample can be a sample of multiple medical images with different contrasts. The water distribution image sample, fat distribution image sample, in-phase image sample, and opposed-phase image sample are respectively samples of the water distribution image, fat distribution image, in-phase image, and opposed-phase image. The image sample label can be an annotation of the set tissue boundary and name in the multi-contrast image sample.

[0081] In specific implementation, the terminal obtains the multi-contrast image sample of the target part and the image sample label corresponding to each multi-contrast image sample, inputs the multi-contrast image sample and the image sample label into the target recognition model to be trained. The target recognition model to be trained recognizes the multi-contrast image sample to obtain the recognition result of the multi-contrast image sample, compares the recognition result with the image sample label. If the gap between the two does not exceed the preset threshold or reaches the preset number of iterations, the target recognition model at this time is used as the trained target recognition model; otherwise, if the gap between the two exceeds the preset threshold or does not reach the preset number of iterations, the model parameters of the target recognition model are adjusted, and the above process is repeated until the gap between the recognition result and the image sample label does not exceed the preset threshold or reaches the preset number of iterations, and the target recognition model at this time is used as the trained target recognition model.

[0082] Figure 6 A flowchart of the training process of a target recognition model is provided. According to Figure 6, it is possible to obtain three-dimensional water distribution images, fat distribution images, in-phase images, and opposed-phase images of the same target part or multiple different target parts in one scan, as three-dimensional multi-contrast image samples. Mark the names and boundaries of the muscles in the three-dimensional multi-contrast image samples to obtain image sample labels. Input the three-dimensional multi-contrast image samples and image sample labels into the target recognition model to be trained, and train the target recognition model to be trained. Specifically, the target recognition model to be trained can compare the muscle names and boundaries recognized from the three-dimensional multi-contrast image samples with the image sample labels, and adjust the parameters of the target recognition model according to the comparison results until the preset iteration termination condition is met, obtaining the trained target recognition model.

[0083] In this embodiment, by obtaining multi-contrast image samples of the target part; the multi-contrast image samples include water distribution image samples, fat distribution image samples, in-phase image samples, and opposed-phase image samples; training the target recognition model to be trained according to the multi-contrast image samples and the corresponding image sample labels of the multi-contrast image samples to obtain the trained target recognition model, the target recognition model can be trained to determine the recognition results of the set tissue through the target recognition model, improving the recognition efficiency of the set tissue.

[0084] In one embodiment, after the step of obtaining the multi-contrast image samples of the target part, it can specifically further include: separately segmenting the set tissue in the water distribution image sample, fat distribution image sample, in-phase image sample, and opposed-phase image sample to obtain the first sample segmentation result, second sample segmentation result, third sample segmentation result, and fourth sample segmentation result of the set tissue; fusing the first sample segmentation result, second sample segmentation result, third sample segmentation result, and fourth sample segmentation result to obtain the fused sample segmentation result; obtaining the image sample label according to the fused sample segmentation result.

[0085] In specific implementation, the terminal can segment the set tissue in the water distribution image sample to obtain the first sample segmentation result, segment the set tissue in the fat distribution image sample to obtain the second sample segmentation result, segment the set tissue in the in-phase image sample to obtain the third sample segmentation result, segment the set tissue in the opposed-phase image sample to obtain the fourth sample segmentation result, fuse the first sample segmentation result, second sample segmentation result, third sample segmentation result, and fourth sample segmentation result to obtain the fused sample segmentation result, and use the fused sample segmentation result as the image sample label of the multi-contrast image sample.

[0086] In this embodiment, by separately segmenting the specified tissues in the water distribution image sample, fat distribution image sample, in-phase image sample, and opposed-phase image sample, the first sample segmentation result, second sample segmentation result, third sample segmentation result, and fourth sample segmentation result of the specified tissue are obtained; the first sample segmentation result, second sample segmentation result, third sample segmentation result, and fourth sample segmentation result are fused to obtain the fused sample segmentation result; according to the fused sample segmentation result, an image sample label is obtained. By fusing the sample segmentation results of the water distribution image sample, fat distribution image sample, in-phase image sample, and opposed-phase image sample, the accuracy of the image sample label can be improved, and further the accuracy of the trained target recognition model can be improved.

[0087] In one embodiment, the step of fusing the first sample segmentation result, second sample segmentation result, third sample segmentation result, and fourth sample segmentation result to obtain the fused sample segmentation result may specifically include: determining an initial sample segmentation result from the first sample segmentation result, second sample segmentation result, third sample segmentation result, and fourth sample segmentation result, and determining the sample segmentation results other than the initial sample segmentation result as adjustment sample segmentation results; respectively adjusting the initial sample segmentation result according to each adjustment sample segmentation result to obtain the fused sample segmentation result.

[0088] Among them, the initial sample segmentation result may be any one of the first sample segmentation result, second sample segmentation result, third sample segmentation result, and fourth sample segmentation result. The adjustment sample segmentation result may be the first sample segmentation result, second sample segmentation result, third sample segmentation result, or fourth sample segmentation result other than the initial sample segmentation result.

[0089] In specific implementation, the terminal may first determine an initial sample segmentation result from the first sample segmentation result, second sample segmentation result, third sample segmentation result, and fourth sample segmentation result, and determine the remaining three sample segmentation results as adjustment sample segmentation results, and then use each adjustment sample segmentation result to adjust the initial sample segmentation result respectively, and finally obtain the fused sample segmentation result.

[0090] For example, the first sample segmentation result of the water distribution image sample may be determined as the initial sample segmentation result, first adjusted according to the second sample segmentation result of the fat distribution image sample to obtain a preliminarily adjusted sample segmentation result, then adjusted according to the third sample segmentation result of the in-phase image sample to the preliminarily adjusted sample segmentation result to obtain a re-adjusted sample segmentation result, and then adjusted according to the fourth sample segmentation result of the opposed-phase image sample to the re-adjusted sample segmentation result to obtain the fused sample segmentation result.

[0091] In this embodiment, by determining an initial sample segmentation result from the first sample segmentation result, the second sample segmentation result, the third sample segmentation result, and the fourth sample segmentation result, the sample segmentation results other than the initial sample segmentation result are determined as adjusted sample segmentation results; the initial sample segmentation result is adjusted according to each adjusted sample segmentation result to obtain a fused sample segmentation result, which can fuse the sample segmentation results of the water distribution image sample, the fat distribution image sample, the in-phase image sample, and the opposed-phase image sample, improve the accuracy of the image sample labels, and further improve the accuracy of the trained target recognition model.

[0092] In one embodiment, step S130 above may specifically include: obtaining a change curve of tissue parameters of a set tissue according to the recognition result of the set tissue corresponding to at least one target moment; and displaying the change curve of the tissue parameters.

[0093] Wherein, the target moment may be the moment targeted for quantitative analysis.

[0094] Wherein, the tissue parameters may be the parameters examined for the set tissue.

[0095] In specific implementation, the terminal may, at at least one target moment, obtain the recognition result of the set tissue of the target part according to the multi-contrast image of the target part, determine the tissue parameters of the set tissue according to the recognition result, use each target moment as the time axis, plot the tissue parameters of the set tissue corresponding to each target moment, generate a change curve of the tissue parameters of the set tissue at each target moment, and display the generated change curve.

[0096] Figure 7 A flowchart of the recognition process of a target recognition model is provided. According to Figure 7 , three-dimensional multi-contrast images of the target part can be obtained at three moments: six months ago, three months ago, and now, and the muscle segmentation results corresponding to the three-dimensional multi-contrast images can be recognized through the trained target recognition model. According to the muscle segmentation results, the muscle volume is statistically calculated to obtain the muscle volumes corresponding to the target part six months ago, three months ago, and now respectively, and then a change curve of the muscle volume six months ago, three months ago, and now is generated.

[0097] In this embodiment, by obtaining a change curve of tissue parameters of a set tissue according to the recognition result of the set tissue corresponding to at least one target moment; and displaying the change curve of the tissue parameters, the change curve of the tissue parameters of the set tissue over time can be visually displayed, which is convenient for observing the change of the tissue parameters.

[0098] In one embodiment, the tissue parameter includes volume; the step of obtaining the change curve of the tissue parameter of the set tissue according to the recognition result of the set tissue corresponding to at least one target time may specifically include: determining the volume of the set tissue at each target time according to the recognition result of the set tissue; obtaining the volume change curve of the set tissue according to the volume at each target time.

[0099] In specific implementation, the terminal may, at each target time, determine the volume of the set tissue according to the recognition result of the set tissue, and draw the volume of the set tissue corresponding to each target time with each target time as the time axis to obtain the volume change curve of the set tissue.

[0100] Figure 8 A schematic diagram of the muscle volume change curve is provided. According to Figure 8 , the muscle volume six months ago, three months ago and now can be determined according to the muscle recognition result, and the muscle volume change curve can be generated. Among them, the volume of muscle C increased significantly from six months ago to three months ago, and the volume recovered to the level of six months ago from three months to now. The volume of muscle D has not changed significantly in the past six months.

[0101] In this embodiment, by determining the volume of the set tissue at each target time according to the recognition result of the set tissue, and obtaining the volume change curve of the set tissue according to the volume at each target time, the change curve of the volume of the set tissue over time can be intuitively displayed, which is convenient for observing the volume change of the set tissue at the target site.

[0102] In one embodiment, the tissue parameter further includes volume ratio; the step of obtaining the change curve of the tissue parameter of the set tissue according to the recognition result of the set tissue corresponding to at least one target time may specifically further include: determining the volume ratio of the set tissue in the target site at each target time according to the recognition result of the set tissue; obtaining the volume ratio change curve of the set tissue according to the volume ratio at each target time.

[0103] In specific implementation, the terminal may, at each target time, determine the volume of the set tissue according to the recognition result of the set tissue, and determine the proportion of the volume in the target site to obtain the volume ratio, and draw the volume ratio of the set tissue in the target site corresponding to each target time with each target time as the time axis to obtain the volume ratio change curve of the set tissue.

[0104] For example, according to the muscle recognition result, the proportion of the volume of the shoulder muscle in the entire shoulder six months ago, three months ago and now can be determined respectively to obtain the volume ratio of the shoulder muscle, and the change curve of the volume ratio six months ago, three months ago and now can be drawn to obtain the volume ratio change curve of the shoulder muscle.

[0105] In this embodiment, according to the recognition result of the set tissue, the volume ratio of the set tissue at the target site at each target time is determined; according to the volume ratio at each target time, the volume ratio change curve of the set tissue is obtained, which can intuitively display the change curve of the ratio of the set tissue at the target site over time, facilitating the observation of the ratio change of the set tissue at the target site.

[0106] To facilitate those skilled in the art to deeply understand the embodiments of the present application, a specific example will be described below.

[0107] To help fitness enthusiasts, rehabilitation doctors, and rehabilitation training groups intuitively and quantitatively observe which muscles have grown due to exercise, and which muscles have been compensated due to incorrect exercise postures and have grown undesirably, and then accurately correct the exercise methods and movements, the present application proposes a method for automatically evaluating the long-term evolution of muscle states.

[0108] First, to solve the problem of inaccurate thick-layer volume statistics, high-definition three-dimensional scanning is used to display muscle morphology. Figure 9 A schematic diagram of the three-dimensional scanning imaging result is provided. According to Figure 9 , one-time scanning imaging can clearly perform three-dimensional reconstruction and comprehensively present the morphology of tissues such as muscles.

[0109] Second, to solve the problem of inaccurate muscle boundary segmentation, imaging is performed by reconstructing multiple contrast images in one scan. One scan reconstructs water distribution images, fat distribution images, in-phase images, and opposed-phase images. Among them, the water distribution images, fat distribution images, in-phase images, and opposed-phase images have different contrasts, and at the same time, the coordinates are aligned and no registration is required. Combining the water distribution images, fat distribution images, in-phase images, and opposed-phase images for annotation can obtain a more accurate segmentation boundary. Combining with artificial intelligence algorithms can effectively ensure the accuracy of the segmentation results.

[0110] The water-fat separation technique can be used. Based on the characteristic that fat and water are at different resonance frequencies, combined with the three-dimensional gradient echo (GRE) sequence, four contrast images, namely water distribution image, fat distribution image, in-phase image, and opposed-phase image, can be generated in one scan. Among them, the in-phase image is the result of adding water and fat, and the opposed-phase image is the result of subtracting water and fat. The signals of human magnetic resonance mainly come from hydrogen protons in the human body. The precession frequency of hydrogen protons in water is 3.5 ppm (150 Hz / T) faster than that of hydrogen protons in fat. For the hydrogen protons in water and fat in the same pixel, after being excited by a radiofrequency pulse, the transverse magnetization vectors of water and fat are in the same phase. When the radiofrequency pulse stops, because the hydrogen protons in water precess faster than those in fat, after a specific time, when the phase difference between the hydrogen protons in water and those in fat is 180 degrees, their macroscopic transverse magnetization vectors cancel each other out (Mxy = 0), and the signal detected by magnetic resonance is the difference after subtracting the water and fat signals. At this time, the image is the opposed-phase image. The opposed-phase image will have an obvious edge-delineation effect visually. Because usually the signal of muscle tissue mainly comes from water molecules, and the space around the muscle is filled with adipose tissue, so on the opposed-phase image, the signal drops of both muscle and the surrounding adipose tissue are not obvious. However, at the junction of the two, there are both muscle (water molecules) and fat, so the signal on the opposed-phase image is significantly reduced, thus resulting in the edge-delineation effect. The edge-delineation effect helps to better present the tissue boundary.

[0111] As Figure 3 and 4 shown, the fat distribution image (upper right) and the opposed-phase image (lower right) show the clearest boundaries of muscles and bones. The opposed-phase image shows the clearest demarcation between two muscles. The water distribution image (upper left) is the most sensitive to the contour of the lesion, and the composition of the lesion needs to be analyzed in combination with the in-phase image (lower left). Therefore, after combining the water distribution image, fat distribution image, in-phase image, and opposed-phase image, the obtained muscle segmentation boundary and disease diagnosis result are the most accurate.

[0112] Specifically, the contour of the muscle can be initially outlined on one of the four contrast images, namely the water distribution image, fat distribution image, in-phase image, and opposed-phase image, and then the initially outlined contour is mapped to the other three contrast images, and fine adjustments are made on the other three contrast images one by one until the muscle boundaries on the four contrast images are consistent. This method can also be applied to the determination of the image sample labels in the training stage of the target recognition model, so that the gold standard input to the target recognition model is unique and accurate, thus ensuring the accuracy of the segmentation contour output after the model training.

[0113] Finally, to solve the problem of intuitively understanding the differences at different times, the scanning results at different time points can also be automatically compared to visually present the volume and morphological change trends of the target muscle at different time points.

[0114] The muscle segmentation results obtained by combining multiple images of the water distribution image, fat distribution image, in-phase image, and opposed-phase image are as Figure 5 shown. Different muscles can be represented by different colors or grayscales, the muscle segmentation boundaries can be marked on the water distribution image, the muscle volume can also be statistically calculated based on the muscle segmentation boundaries, and the muscle volumes obtained at multiple moments can be compared along the time axis.

[0115] It should be noted that the above method is not limited to the comparison based on muscles, but can also be the comparison of fat volume or the comparison of the volume of specific organs; the scanning method is not limited to a specific sequence, but can also be the combination of multiple sequences. For example, T1WI (longitudinal relaxation time weighted imaging), T2WI (transverse relaxation time weighted imaging), DWI (Diffusion Weighted Imaging), contrast agent enhancement technology, etc.; and the scanning device is not limited to magnetic resonance, but can also be the combination of multiple modalities. For example, the combination of MR (Magnetic Resonance) and CT (Computed Tomography), or the combination of MR and PET (Positron Emission Computed Tomography).

[0116] The above method for automatically evaluating the long-term evolution of muscle status can visually present the three-dimensional morphology of high-definition muscles. Instead of estimating the muscle status through the outer skin, it can segment the muscle boundaries, calculate the volume sizes of different muscles, and compare the muscle examination results at different times to visually present the change trends of the target muscle, which is convenient for fitness enthusiasts, rehabilitation doctors, and rehabilitation training groups to intuitively and quantitatively see which muscles have grown due to exercise and which muscles have had abnormal growth due to compensatory effects of incorrect exercise postures, so as to accurately correct the exercise methods and movements.

[0117] In one embodiment, as Figure 10 shown, a medical information analysis method applicable to medical images is provided. Taking the application of this method to a terminal as an example, it includes the following steps:

[0118] Step S201, obtaining multi-contrast image samples of the target part; the multi-contrast image samples include water distribution image samples, fat distribution image samples, in-phase image samples, and opposed-phase image samples;

[0119] Step S202: Segment the specified tissue in the water distribution image sample, fat distribution image sample, in-phase image sample, and opposed-phase image sample respectively to obtain the first sample segmentation result, second sample segmentation result, third sample segmentation result, and fourth sample segmentation result of the specified tissue;

[0120] Step S203: Fuse the first sample segmentation result, second sample segmentation result, third sample segmentation result, and fourth sample segmentation result to obtain the fused sample segmentation result, and obtain the image sample label according to the fused sample segmentation result;

[0121] Step S204: Train the target recognition model to be trained according to the multi-contrast image sample and the corresponding image sample label of the multi-contrast image sample to obtain the trained target recognition model;

[0122] Step S205: Obtain the multi-contrast image of the target part; the multi-contrast image includes a water distribution image, a fat distribution image, an in-phase image, and an opposed-phase image;

[0123] Step S206: Input the water distribution image, fat distribution image, in-phase image, and opposed-phase image into the trained target recognition model to obtain the recognition result of the specified tissue of the target part;

[0124] Step S207: Display the quantitative analysis result of the specified tissue of the target part in a visual form according to the recognition result of the specified tissue.

[0125] In specific implementation, the multi-contrast image sample of the target part collected by the medical imaging device can be input into the terminal. The terminal segments the multi-contrast image sample to obtain the first sample segmentation result, second sample segmentation result, third sample segmentation result, and fourth sample segmentation result, fuses the first sample segmentation result, second sample segmentation result, third sample segmentation result, and fourth sample segmentation result, and uses the obtained fused sample segmentation result as the image sample label corresponding to the multi-contrast image sample. Train the target recognition model to be trained according to the multi-contrast image sample and the image sample label to obtain the trained target recognition model. Collect the multi-contrast image of the target part and input it into the terminal. The terminal inputs the multi-contrast image into the trained target recognition model to recognize the specified tissue of the target part, obtains the recognition result of the specified tissue, determines the tissue parameters of the specified tissue according to the recognition result, and displays the tissue parameters in the form of an icon to obtain the quantitative analysis result of the specified tissue.

[0126] The above-mentioned medical information analysis method applicable to medical images obtains multi-contrast image samples of the target part, respectively segments the set tissues in the water distribution image sample, fat distribution image sample, in-phase image sample and opposed-phase image sample to obtain the first sample segmentation result, second sample segmentation result, third sample segmentation result and fourth sample segmentation result of the set tissues, fuses the first sample segmentation result, second sample segmentation result, third sample segmentation result and fourth sample segmentation result to obtain the fused sample segmentation result, obtains the image sample label according to the fused sample segmentation result, trains the target recognition model to be trained according to the multi-contrast image sample and the image sample label corresponding to the multi-contrast image sample to obtain the trained target recognition model, obtains the multi-contrast image of the target part, inputs the water distribution image, fat distribution image, in-phase image and opposed-phase image into the trained target recognition model to obtain the recognition result of the set tissues of the target part, and displays the quantitative analysis result of the set tissues of the target part in a visual form according to the recognition result of the set tissues; it can identify set tissues such as total body fat, muscle, bone, liver, uterus, prostate, etc. based on the multi-contrast image, improve the accuracy of the recognition result of the set tissues, and further improve the accuracy of medical information analysis. Visualizing the quantitative analysis result of the set tissues also increases the readability of the medical information analysis result.

[0127] It should be understood that although each step in the flowcharts involved in the above-described embodiments is shown in sequence according to the indication of the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless there is a clear indication in this article, there is no strict order restriction for the execution of these steps, and these steps can be executed in other orders. Moreover, at least a part of the steps in the flowcharts involved in the above-described embodiments may include multiple steps or multiple stages. These steps or stages are not necessarily executed at the same moment, but can be executed at different moments. The execution order of these steps or stages is not necessarily sequential, but can be executed alternately or in turn with at least a part of other steps or steps or stages in other steps.

[0128] Based on the same inventive concept, the embodiments of the present application also provide a medical information analysis device applicable to medical images for implementing the above-mentioned medical information analysis method applicable to medical images. The solution provided by this device for solving problems is similar to the solution described in the above method. Therefore, the specific limitations in one or more embodiments of the medical information analysis device applicable to medical images provided below can refer to the limitations on the medical information analysis method applicable to medical images in the above text, and will not be repeated here.

[0129] In one embodiment, as Figure 11As shown, a medical information analysis device applicable to medical images is provided, including: an acquisition module 310, an identification module 320, and a determination module 330, where:

[0130] The acquisition module 310 is configured to acquire multi-contrast images of a target part; the multi-contrast images include a water distribution image, a fat distribution image, an in-phase image, and an opposed-phase image;

[0131] The identification module 320 is configured to input the water distribution image, the fat distribution image, the in-phase image, and the opposed-phase image into a trained target recognition model to obtain an identification result of a set tissue of the target part;

[0132] The display module 330 is configured to display a quantitative analysis result of the set tissue of the target part in a visual form according to the identification result of the set tissue.

[0133] In one embodiment, the above-mentioned medical information analysis device applicable to medical images further includes:

[0134] A sample acquisition module, configured to acquire multi-contrast image samples of the target part; the multi-contrast image samples include a water distribution image sample, a fat distribution image sample, an in-phase image sample, and an opposed-phase image sample;

[0135] A model training module, configured to train a target recognition model to be trained according to the multi-contrast image samples and the image sample labels corresponding to the multi-contrast image samples to obtain the trained target recognition model.

[0136] In one embodiment, the above-mentioned medical information analysis device applicable to medical images further includes:

[0137] A sample segmentation module, configured to respectively segment a set tissue in the water distribution image sample, the fat distribution image sample, the in-phase image sample, and the opposed-phase image sample to obtain a first sample segmentation result, a second sample segmentation result, a third sample segmentation result, and a fourth sample segmentation result of the set tissue;

[0138] A sample fusion module, configured to fuse the first sample segmentation result, the second sample segmentation result, the third sample segmentation result, and the fourth sample segmentation result to obtain a fused sample segmentation result;

[0139] A sample label module, configured to obtain the image sample label according to the fused sample segmentation result.

[0140] In one embodiment, the above sample fusion module is further configured to determine an initial sample segmentation result from the first sample segmentation result, the second sample segmentation result, the third sample segmentation result, and the fourth sample segmentation result, and determine the sample segmentation results other than the initial sample segmentation result as adjusted sample segmentation results; and adjust the initial sample segmentation result according to each of the adjusted sample segmentation results to obtain the fused sample segmentation result.

[0141] In one embodiment, the above determination module 330 is further configured to obtain a change curve of the tissue parameters of the set tissue according to the recognition result of the set tissue corresponding to at least one target time; and display the change curve of the tissue parameters.

[0142] In one embodiment, the above determination module 330 is further configured to determine the volume of the set tissue at each of the target times according to the recognition result of the set tissue; and obtain a volume change curve of the set tissue according to the volumes at each of the target times.

[0143] In one embodiment, the above determination module 330 is further configured to determine the volume ratio of the set tissue at the target site at each of the target times according to the recognition result of the set tissue; and obtain a volume ratio change curve of the set tissue according to the volume ratios at each of the target times.

[0144] Each module in the above medical information analysis device applicable to medical images can be implemented in whole or in part by software, hardware, and their combination. Each of the above modules can be embedded in the processor of the computer device in hardware form or independent of it, or stored in the memory of the computer device in software form, so that the processor can call and execute the operations corresponding to each of the above modules.

[0145] In one embodiment, a computer device is provided. The computer device may be a terminal, and its internal structure diagram may be as Figure 12As shown in the figure. The computer device includes a processor, a memory, an input / output interface, a communication interface, a display unit, and an input device. Among them, the processor, the memory, and the input / output interface are connected through a system bus, and the communication interface, the display unit, and the input device are connected to the system bus through the input / output interface. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The input / output interface of the computer device is used to exchange information between the processor and external devices. The communication interface of the computer device is used to communicate with external terminals in a wired or wireless manner, and the wireless manner can be achieved through WIFI, a mobile cellular network, NFC (Near Field Communication), or other technologies. When the computer program is executed by the processor, it implements a medical information analysis method applicable to medical images. The display unit of the computer device is used to form a visually visible picture, which can be a display screen, a projection device, or a virtual reality imaging device. The display screen can be a liquid crystal display screen or an electronic ink display screen. The input device of the computer device can be a touch layer covering the display screen, or a button, a trackball, or a touchpad provided on the housing of the computer device, or an external keyboard, touchpad, or mouse, etc.

[0146] Those skilled in the art can understand that Figure 12 the structure shown in the figure is only a block diagram of some structures related to the solution of this application, and does not constitute a limitation on the computer device to which the solution of this application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine some components, or have different component arrangements.

[0147] In one embodiment, a computer device is further provided, including a memory and a processor. A computer program is stored in the memory, and when the processor executes the computer program, the steps in the above method embodiments are implemented.

[0148] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored, and when the computer program is executed by the processor, the steps in the above method embodiments are implemented.

[0149] In one embodiment, a computer program product is provided, including a computer program, and when the computer program is executed by the processor, the steps in the above method embodiments are implemented.

[0150] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use, and processing of relevant data need to comply with the relevant laws, regulations, and standards of relevant countries and regions.

[0151] Those of ordinary skill in the art can understand that all or part of the processes in the methods of the above embodiments can be completed by instructing relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above methods. Among them, any reference to a memory, database, or other medium used in the embodiments provided in this application can include at least one of non-volatile and volatile memories. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetoresistive random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM), etc. The databases involved in the embodiments provided in this application can include at least one of relational databases and non-relational databases. Non-relational databases can include distributed databases based on blockchain, etc., and are not limited thereto. The processors involved in the embodiments provided in this application can be general-purpose processors, central processors, graphics processors, digital signal processors, programmable logic devices, data processing logics based on quantum computing, etc., and are not limited thereto.

[0152] The technical features of the above embodiments can be combined arbitrarily. For the sake of brevity of description, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, it should be considered to be within the scope described in this specification.

[0153] The above-described embodiments merely represent several implementation manners of the present application. The description thereof is relatively specific and detailed, but it should not be construed as a limitation on the patent scope of the present application. It should be noted that for those of ordinary skill in the art, without departing from the concept of the present application, several modifications and improvements can still be made, and these all fall within the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the appended claims.

Claims

1. A medical information analysis method applicable to medical images, characterized in that, The method includes: Obtaining multi-contrast images of a target part; the multi-contrast images include a water distribution image, a fat distribution image, an in-phase image, and an opposed-phase image; Inputting the water distribution image, the fat distribution image, the in-phase image, and the opposed-phase image into a trained target recognition model to obtain a recognition result of a set tissue of the target part; According to the recognition result of the set tissue, visually displaying a quantitative analysis result of the set tissue of the target part.

2. The method according to claim 1, characterized in that, Before obtaining the multi-contrast images of the target part, it further includes: Obtaining multi-contrast image samples of the target part; the multi-contrast image samples include a water distribution image sample, a fat distribution image sample, an in-phase image sample, and an opposed-phase image sample; Training a target recognition model to be trained according to the multi-contrast image samples and the image sample labels corresponding to the multi-contrast image samples to obtain the trained target recognition model.

3. The method according to claim 2, characterized in that, After obtaining the multi-contrast image samples of the target part, it further includes: Respectively segmenting the set tissue in the water distribution image sample, the fat distribution image sample, the in-phase image sample, and the opposed-phase image sample to obtain a first sample segmentation result, a second sample segmentation result, a third sample segmentation result, and a fourth sample segmentation result of the set tissue; Fusing the first sample segmentation result, the second sample segmentation result, the third sample segmentation result, and the fourth sample segmentation result to obtain a fused sample segmentation result; According to the fused sample segmentation result, obtaining the image sample label.

4. The method according to claim 3, characterized in that, The fusing the first sample segmentation result, the second sample segmentation result, the third sample segmentation result, and the fourth sample segmentation result to obtain a fused sample segmentation result includes: Determining an initial sample segmentation result from the first sample segmentation result, the second sample segmentation result, the third sample segmentation result, and the fourth sample segmentation result, and determining the sample segmentation results other than the initial sample segmentation result as adjusted sample segmentation results; Respectively adjusting the initial sample segmentation result according to each of the adjusted sample segmentation results to obtain the fused sample segmentation result.

5. The method according to claim 1, characterized in that, The visually displaying a quantitative analysis result of the set tissue of the target part according to the recognition result of the set tissue includes: According to the recognition result of the set tissue corresponding to at least one target time, obtaining a change curve of tissue parameters of the set tissue; Displaying the change curve of the tissue parameters.

6. The method according to claim 5, characterized in that, The tissue parameters include volume; the obtaining a change curve of tissue parameters of the set tissue according to the recognition result of the set tissue corresponding to at least one target time includes: According to the recognition result of the set tissue, determining the volume of the set tissue at each of the target times; According to the volumes at each of the target times, obtaining a volume change curve of the set tissue.

7. The method according to claim 5, characterized in that, The tissue parameters further include volume proportion; obtaining the change curve of the tissue parameters of the set tissue according to the recognition result of the set tissue corresponding to at least one target time further includes: Determining the volume proportion of the set tissue at each target time in the target part according to the recognition result of the set tissue; Obtaining the change curve of the volume proportion of the set tissue according to the volume proportion at each target time.

8. A medical information analysis device applicable to medical images, characterized in that, The device includes: An acquisition module, configured to acquire multi-contrast images of a target part; the multi-contrast images include a water distribution image, a fat distribution image, an in-phase image, and an opposed-phase image; An identification module, configured to input the water distribution image, the fat distribution image, the in-phase image, and the opposed-phase image into a trained target recognition model to obtain the recognition result of the set tissue of the target part; A display module, configured to visually display the quantitative analysis result of the set tissue of the target part according to the recognition result of the set tissue.

9. A computer device, comprising a memory and a processor, the memory storing a computer program, characterized in that, When the processor executes the computer program, the steps of the method according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium, on which a computer program is stored, characterized in that, When the computer program is executed by the processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images