Rheumatic mitral valve operation success probability prediction method, device and equipment

Abstract

The invention relates to the technical field of medical information processing, and discloses a rheumatic mitral valve operation success probability prediction method, device and equipment, and the method comprises the steps: obtaining target information of a mitral valve of a target object, the target information at least comprising mitral valve calcification information, mitral valve leaflet fiber thickening information and papillary muscle symmetry information; the target information is obtained based on a cardiac computed tomography image of a target object; inputting the target information into a pre-constructed and trained operation success probability prediction model for prediction; the operation success probability prediction model is used for predicting the probability of successful repair of the mitral valve; and obtaining the operation success probability output by the operation success probability prediction model, wherein the operation success probability is used for guiding the selection of the mitral valve operation strategy of the target object. According to the method, the probability of good repair of the mitral valve of the RMD patient can be accurately predicted, so that surgical treatment prompts and guidance are provided, and the optimal surgical scheme can be selected according to lesion conditions in a personalized manner.

Description

Technical Field

[0001] The present invention relates to the technical field of medical information processing, and in particular to a method, device and equipment for predicting the success probability of rheumatic mitral valve surgery. Background Art

[0002] Rheumatic mitral disease (RMD) is a sequelae of acute rheumatic fever and the result of mitral valve damage caused by an abnormal immune response to group A streptococcal infection. Patients with severe RMD often require interventions, including percutaneous mitral balloon commissurotomy (PMBC) and surgical valve replacement or repair.

[0003] The selection of the best intervention strategy is inseparable from the accurate preoperative assessment of the patient's valve structure and the extent of the disease. Currently, the standard examination method for RMD patients is transthoracic echocardiography. The semi-quantitative scores of Wilkins and Cormier based on echocardiography are the two most widely used scoring schemes for assessing the extent of disease in RMD patients. However, they are mainly designed for the screening needs of patients with PMBC and have great limitations in guiding surgical intervention. For a large number of RMD patients with contraindications to PMBC in clinical practice, they can only be treated through surgical valve repair or replacement surgery. Studies have found that repair surgery in RMD patients with appropriate lesions can achieve long-term efficacy better than mitral valve replacement. However, the relevant technology lacks an objective and effective preoperative evaluation scheme to accurately guide the choice of surgical strategy. How to select RMD patients suitable for repair has always troubled surgeons. Summary of the Invention

[0004] In view of this, the present invention provides a method, device and equipment for predicting the success probability of rheumatic mitral valve surgery to solve the problem of being unable to objectively and effectively evaluate the appropriate surgical strategy for RMD patients.

[0005] In a first aspect, the present invention provides a method for predicting the success probability of rheumatic mitral valve surgery, the method comprising:

[0006] acquiring target information of a mitral valve of a target subject, the target information including at least mitral valve calcification information, mitral valve leaflet fiber thickening information, and papillary muscle symmetry information; the target information is measured based on a reconstructed three-dimensional image of the mitral valve, the three-dimensional image of the mitral valve being reconstructed based on a cardiac computed tomography image of the target subject;

[0007] Inputting the target information into a pre-built and trained surgical success probability prediction model for prediction; the surgical success probability prediction model is used to predict the probability of successful mitral valve repair;

[0008] The surgical success probability output by the surgical success probability prediction model is obtained, and the surgical success probability is used to guide the selection of a mitral valve surgical strategy for the target subject.

[0009] In an optional embodiment, the mitral valve calcification information includes at least one of the following:

[0010] Agatston score of mitral valve calcification area;

[0011] calcification volume integral of the mitral valve calcification area;

[0012] Calcification mass integral of the mitral valve calcification area;

[0013] Location information of mitral valve calcification areas;

[0014] The degree of mitral valve leaflet calcification invasion;

[0015] Thickness of residual leaflets in the mitral valve calcification area.

[0016] In an optional embodiment, the method for predicting the success probability of rheumatic mitral valve surgery further includes:

[0017] Based on the degree of mitral valve leaflet calcification and / or the thickness of the residual leaflet at the mitral valve calcification site, the probability of mitral valve leaflet rupture when surgical repair is performed to remove the calcification on the leaflet is determined.

[0018] In an optional embodiment, the mitral valve leaflet fiber thickening information includes at least one of the following: thickness information of the A1 area of ​​the anterior leaflet; thickness information of the A2 area of ​​the anterior leaflet; thickness information of the A3 area of ​​the anterior leaflet; thickness information of the P1 area of ​​the posterior leaflet; thickness information of the P2 area of ​​the posterior leaflet; thickness information of the P3 area of ​​the posterior leaflet;

[0019] The thickness information of the A1 region of the anterior lobe includes: thickness information of the zona pellucida corresponding to the A1 region of the anterior lobe, thickness information of the zona sclera corresponding to the A1 region of the anterior lobe, and overall thickness information of the A1 region of the anterior lobe;

[0020] The thickness information of the A2 area of ​​the anterior leaflet includes: the thickness information of the zona pellucida corresponding to the A2 area of ​​the anterior leaflet, the thickness information of the zona crassa corresponding to the A2 area of ​​the anterior leaflet, and the overall thickness information of the A2 area of ​​the anterior leaflet;

[0021] The thickness information of the A3 area of ​​the anterior leaflet includes: the thickness information of the zona pellucida corresponding to the A3 area of ​​the anterior leaflet, the thickness information of the zona crassa corresponding to the A3 area of ​​the anterior leaflet, and the overall thickness information of the A3 area of ​​the anterior leaflet;

[0022] The thickness information of the P1 region of the posterior leaflet includes: the thickness information of the rough zone portion corresponding to the P1 region of the posterior leaflet and the overall thickness information of the P1 region of the posterior leaflet;

[0023] The thickness information of the P2 region of the posterior leaflet includes: thickness information of the rough zone portion corresponding to the P2 region of the posterior leaflet and overall thickness information of the P2 region of the posterior leaflet;

[0024] The thickness information of the P3 region of the posterior lobe includes: thickness information of the rough zone portion corresponding to the P3 region of the posterior lobe and overall thickness information of the P3 region of the posterior lobe.

[0025] In an optional implementation, the target information further includes at least one of the following:

[0026] mitral valve leaflet contracture length;

[0027] residual chordae length;

[0028] Papillary muscle fusion information;

[0029] The major and minor diameters of the annulus.

[0030] In an optional embodiment, acquiring target information of the mitral valve of the target object includes:

[0031] Using a target information recognition model, recognizing a cardiac computed tomography image of the target object;

[0032] determining the target information based on an output of the target information recognition model;

[0033] Among them, the target information recognition model is a multi-task recognition model, which has a shared feature network and multiple branch output networks. The shared feature network is used to extract the feature information of the cardiac computed tomography image, and the multiple branch output networks respectively predict different target information based on the feature information output by the shared feature network.

[0034] In an optional embodiment, the surgery success probability prediction model is an uncertainty-aware probability model, which includes an input layer, a feature extractor, an uncertainty estimation module, and an output layer;

[0035] Among them, the input layer is used to receive the target information; the feature extractor is a Transformer network, which is used to extract features from the target information to obtain corresponding feature information; the uncertainty estimation module includes a Bayesian neural network, which is used to output an uncertain estimation result based on the feature information; the output layer is used to generate a prediction result of the probability of success of the operation based on the feature information and the uncertain estimation result, and output the predicted probability of success of the operation and the uncertain estimation result.

[0036] In an optional embodiment, inputting the target information into a pre-built and trained surgery success probability prediction model for prediction includes:

[0037] obtaining coronary artery disease information and / or pulmonary artery pressure information of the target subject;

[0038] The target information, as well as the coronary artery disease information and / or the pulmonary artery pressure information, are input into the surgery success probability prediction model for prediction.

[0039] In a second aspect, the present invention provides a device for predicting the success probability of rheumatic mitral valve surgery, the device comprising:

[0040] a target information acquisition module, configured to acquire target information of the mitral valve of a target subject, the target information including at least mitral valve calcification information, mitral valve leaflet fiber thickening information, and papillary muscle symmetry information; the target information is measured based on a reconstructed three-dimensional image of the mitral valve, the three-dimensional image of the mitral valve being reconstructed based on a cardiac computed tomography image of the target subject;

[0041] A prediction module is used to input the target information into a pre-built and trained surgery success probability prediction model for prediction; the surgery success probability prediction model is used to predict the probability of successful mitral valve repair;

[0042] The operation success probability acquisition module is used to obtain the operation success probability output by the operation success probability prediction model, and the operation success probability is used to guide the selection of the mitral valve surgery strategy for the target object.

[0043] In a third aspect, the present invention provides a medical device comprising: a memory and a processor, the memory and the processor being communicatively connected to each other, the memory storing computer instructions, and the processor executing the computer instructions to execute the method for predicting the success probability of rheumatic mitral valve surgery according to the first aspect or any corresponding embodiment thereof.

[0044] In a fourth aspect, the present invention provides a computer-readable storage medium having computer instructions stored thereon, the computer instructions being used to enable a computer to execute the method for predicting the success probability of rheumatic mitral valve surgery according to the first aspect or any corresponding embodiment thereof.

[0045] In a fifth aspect, the present invention provides a computer program product comprising computer instructions for causing a computer to execute the method for predicting the success probability of rheumatic mitral valve surgery according to the first aspect or any corresponding embodiment thereof.

[0046] The method, device and equipment for predicting the success probability of rheumatic mitral valve surgery provided by the embodiments of the present invention can locate and determine the mitral valve lesion condition before surgery, thereby accurately assessing the degree of mitral valve lesions in RMD patients (i.e., target subjects) and accurately predicting the probability of good mitral valve repair in RMD patients, thereby providing surgical treatment prompts and guidance, and facilitating personalized selection of the best surgical plan based on the lesion condition.

[0047] The embodiments of the present invention propose for the first time specific evaluation indicators: penetrating calcification, papillary muscle symmetry, residual chordae tendineae length, etc. Statistical results show that these factors significantly affect the postoperative effect of mitral valve repair and can be used as standard indicators for subsequent evaluation of mitral valve lesions. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] In order to more clearly illustrate the specific embodiments of the present invention or the technical solutions in related technologies, the following briefly introduces the drawings required for use in the specific embodiments or related technical descriptions. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0049] Figure 1 is a flowchart of a method for predicting the success probability of rheumatic mitral valve surgery according to an embodiment of the present invention;

[0050] Figure 2 is an example diagram of the Agatston integral, calcification volume (ie, calcium capacity), and calcification mass of the mitral valve calcification area according to an embodiment of the present invention;

[0051] Figure 3 is a schematic diagram of the location of partial mitral valve calcification according to an embodiment of the present invention;

[0052] Figure 4 is a schematic diagram of the degree of mitral valve leaflet calcification invasion according to an embodiment of the present invention;

[0053] Figure 5 is a schematic diagram of the thickness of the residual leaflet at the mitral valve calcification site according to an embodiment of the present invention;

[0054] Figure 6 is a schematic diagram of a measurement interface for mitral valve leaflet fiber thickening information according to an embodiment of the present invention;

[0055] Figure 7 is a schematic diagram of measuring mitral valve leaflet fiber thickening information according to an embodiment of the present invention;

[0056] Figure 8 is a schematic diagram of the dimensions of the papillary muscle according to an embodiment of the present invention;

[0057] Figure 9 is a schematic diagram of mitral valve leaflet size measurement according to an embodiment of the present invention;

[0058] Figure 10 is a schematic diagram of measuring the length of residual chordae tendineae of a mitral valve according to an embodiment of the present invention;

[0059] Figure 11 is a schematic diagram of papillary muscle fusion according to an embodiment of the present invention;

[0060] Figure 12 is a schematic diagram of measuring the long and short diameters of the valve annulus according to an embodiment of the present invention;

[0061] Figure 13 is a structural block diagram of a device for predicting the success probability of rheumatic mitral valve surgery according to an embodiment of the present invention;

[0062] Figure 14 Schematic diagram of the hardware structure of the medical device according to the embodiment of the present invention. DETAILED DESCRIPTION

[0063] To make the purpose, technical solutions, and advantages of the embodiments of the present invention more clear, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without making creative efforts shall fall within the scope of protection of the present invention.

[0064] The heart is complex, with the mitral and aortic valves having distinct locations and structures. Because the aortic valve is simple and precisely positioned, preoperative assessments of the aortic valve are almost exclusively performed in related technologies, lacking a method for evaluating mitral valve pathology. Therefore, based on extensive prior research and experimentation, the present invention provides an effective and comprehensive method for evaluating mitral valve pathology, offering valuable guidance for the diagnosis and treatment of mitral valve disease.

[0065] According to an embodiment of the present invention, an embodiment of a method for predicting the probability of success of rheumatic mitral valve surgery is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of executable computer instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0066] In this embodiment, a method for predicting the success probability of rheumatic mitral valve surgery is provided, which can be used to predict the probability of a good prognosis of mitral valve repair surgery and can be used in various computer devices, including medical devices. Figure 1 FIG. 1 is a flow chart of a method for predicting the success probability of rheumatic mitral valve surgery according to an embodiment of the present invention. Figure 1 As shown, the process includes the following steps:

[0067] Step S101, obtaining target information of the mitral valve of a target subject (i.e., an RMD patient), wherein the target information includes at least mitral valve calcification information, mitral valve leaflet fiber thickening information, and papillary muscle symmetry information; the target information is measured based on a reconstructed three-dimensional image of the mitral valve, and the three-dimensional image of the mitral valve is reconstructed based on a cardiac computed tomography (CT) image of the target subject.

[0068] In some optional embodiments, the mitral valve calcification information includes at least one of the following:

[0069] Agatston score of mitral valve calcification area;

[0070] The calcification volume integral of the mitral valve calcification area; specifically, the calcification volume (i.e., calcium capacity) can be directly used as the calcification volume integral;

[0071] The calcification mass integral of the mitral valve calcification area; specifically, the calcification mass can be directly used as the calcification mass integral;

[0072] Location information of mitral valve calcification areas;

[0073] The degree of mitral valve leaflet calcification invasion;

[0074] Thickness of residual leaflets in the mitral valve calcification area.

[0075] Specifically, the coronary artery calcification score sequence can be called up from the cardiac CT image, and then based on the calcium calculation analysis (Agatston score) software, the mitral valve calcification position can be located, the calcification score threshold (≥130Hu) can be set, and the mitral valve calcification area can be outlined layer by layer. After the outline is completed, the Agatston score, calcification volume integral and calcification mass integral of the mitral valve calcification area can be automatically calculated. Figure 2This is an example of the calculated Agatston integral, calcification volume (also known as calcium capacity), and calcification mass of the mitral valve calcification area. In other embodiments, automated image processing techniques and deep learning models can be directly used to obtain the Agatston integral, calcification volume (also known as calcium capacity) integral, and calcification mass integral of the mitral valve calcification area to achieve quantification of mitral valve calcification.

[0076] Regarding the location information of the mitral valve calcification area, CT multi-planar reconstruction technology can be used to reconstruct the mitral valve short-axis section to locate the mitral valve annulus calcification; reconstruct the left ventricular outflow tract section and adjust it layer by layer up and down to locate the chordae tendineae and papillary muscle calcification; reconstruct the three-chamber heart and mitral valve short-axis section and adjust it layer by layer up and down to locate the calcification of the anterior leaflet (A1, A2, A3 areas), the posterior leaflet (P1, P2, P3 areas), and the mitral commissure area (anterior commissure (AC), posterior commissure (PC)). The anterior leaflet calcification can be further located as rough band or zona pellucida calcification based on the relative position of the calcification to the valve root and cusp.

[0077] That is to say, the location of the mitral valve calcification area may include the valve ring location, chordae tendineae location, papillary muscle location, anterior leaflet location, posterior leaflet location and commissural location. Calcification at the anterior leaflet location specifically includes rough zone calcification and zona pellucida calcification. Calcification at the commissural location specifically includes anterior commissural calcification and posterior commissural calcification. A schematic diagram of the location of some mitral valve calcifications is shown below. Figure 3 shown.

[0078] Regarding the degree of mitral valve leaflet calcification invasion, when performing the mitral valve leaflet calcification location operation based on the three-chamber heart and left ventricular outflow tract view (as described above), the grayscale value is adjusted to observe the remaining mitral valve leaflet tissue at the calcification attachment site. When the calcification spot is in contact with the blood flow in the cardiac cavity on one side and borders the mitral valve leaflet tissue on the other side, the calcification here is defined as non-penetrating calcification (such as Figure 4 When the upper and lower boundaries of the calcification are in direct contact with the blood flow in the cardiac cavity, the calcification is defined as penetrating calcification (as shown in the right figure); Figure 4 (As shown in the left figure), this indicates that when surgical repair is performed to remove calcification at this location, there is a high probability of mitral valve leaflet rupture, which requires further suturing of the rupture or patch repair.

[0079] In addition, the degree of mitral valve leaflet calcification invasion may further include indication information of whether the valve annulus is involved and indication information of whether the calcification involves the zona pellucida.

[0080] Regarding the thickness of the residual leaflet at the site of mitral valve calcification, when evaluating the degree of invasion of mitral valve leaflet calcification, if the calcification is non-penetrating calcification, the image is magnified, the measurement scale is called up, and the thickness of the remaining mitral valve tissue at the site of calcification attachment is accurately measured, as follows: Figure 5 When the remaining leaflet thickness at the calcified site is too thin, there is still a risk of leaflet perforation.

[0081] The location information of the mitral valve calcification area, the degree of mitral valve leaflet calcification invasion, and the residual leaflet thickness in the mitral valve calcification area all use image recognition technology or deep learning models to automatically identify the mitral valve annulus, chordae tendineae, papillary muscles, anterior leaflet, posterior leaflet and junction area, as well as calcification foci, and then determine the calcification location, the degree of leaflet calcification invasion and the residual thickness.

[0082] In some specific implementations, the method provided by the embodiments of the present invention further includes:

[0083] Based on the degree of mitral valve leaflet calcification and / or the thickness of the residual leaflet at the mitral valve calcification site, the probability of mitral valve leaflet rupture when surgical repair is performed to remove the calcification on the leaflet is determined.

[0084] The probability of mitral valve leaflet rupture is also used to guide the choice of surgical strategy.

[0085] In some optional embodiments, the mitral valve leaflet fiber thickening information includes at least one of the following: thickness information of the A1 area of ​​the anterior leaflet; thickness information of the A2 area of ​​the anterior leaflet; thickness information of the A3 area of ​​the anterior leaflet; thickness information of the P1 area of ​​the posterior leaflet; thickness information of the P2 area of ​​the posterior leaflet; thickness information of the P3 area of ​​the posterior leaflet;

[0086] The thickness information of the A1 region of the anterior lobe includes: thickness information of the zona pellucida corresponding to the A1 region of the anterior lobe, thickness information of the zona sclera corresponding to the A1 region of the anterior lobe, and overall thickness information of the A1 region of the anterior lobe;

[0087] The thickness information of the A2 area of ​​the anterior leaflet includes: the thickness information of the zona pellucida corresponding to the A2 area of ​​the anterior leaflet, the thickness information of the zona crassa corresponding to the A2 area of ​​the anterior leaflet, and the overall thickness information of the A2 area of ​​the anterior leaflet;

[0088] The thickness information of the A3 area of ​​the anterior leaflet includes: the thickness information of the zona pellucida corresponding to the A3 area of ​​the anterior leaflet, the thickness information of the zona crassa corresponding to the A3 area of ​​the anterior leaflet, and the overall thickness information of the A3 area of ​​the anterior leaflet;

[0089] The thickness information of the P1 region of the posterior leaflet includes: the thickness information of the rough zone portion corresponding to the P1 region of the posterior leaflet and the overall thickness information of the P1 region of the posterior leaflet;

[0090] The thickness information of the P2 region of the posterior leaflet includes: thickness information of the rough zone portion corresponding to the P2 region of the posterior leaflet and overall thickness information of the P2 region of the posterior leaflet;

[0091] The thickness information of the P3 region of the posterior lobe includes: thickness information of the rough zone portion corresponding to the P3 region of the posterior lobe and overall thickness information of the P3 region of the posterior lobe.

[0092] Specifically, if Figure 6 As shown, the A1-P1 section, A2-P2 section, and A3-P3 section can be adjusted in the three-chamber heart and left ventricular outflow tract sections, and as shown in Figure 7 The enlarged image shown is used to call up the measuring scale, and the thickness of the rough band and the transparent band of each area of ​​the anterior leaflet and each area of ​​the posterior leaflet are measured respectively. If the observation shows that the thickening degree of the leaflet in different positions is uneven, the thickness of the leaflet at the thickest and thinnest parts is measured respectively to quantify the range of leaflet thickening. For example, the thickness information of the transparent band in the A1 area of ​​the anterior leaflet may include the thickness information of the thickest part and the thinnest part of the transparent band in the A1 area of ​​the anterior leaflet, the thickness information of the rough band in the A1 area of ​​the anterior leaflet may include the thickness information of the thickest part and the thinnest part of the rough band in the A1 area of ​​the anterior leaflet, and the thickness information of the rough band in the P1 area of ​​the posterior leaflet may include the thickness information of the thickest part and the thinnest part of the rough band in the P1 area of ​​the posterior leaflet. If the observation shows that the thickening degree of the leaflet is uniform, the thickness of the standard position can be measured. At the same time, when performing positioning measurement, the abnormal thickening or thinning of the leaflet is positioned and described to provide reference and prompts for surgical repair.

[0093] In some optional embodiments, papillary muscle symmetry information can be evaluated in the following ways:

[0094] Reconstruct the short axis of the ventricle, pull up and down layer by layer and locate the largest cross-sectional area of ​​the anterior and posterior papillary muscles respectively. The upper and lower diameters and the left and right diameters of the papillary muscle cross section are measured here and defined as the long and short diameters of the papillary muscle cross section (such as Figure 8 The standardized calculation method for papillary muscle symmetry is: Papillary muscle symmetry = long diameter of the anterior papillary muscle at its thickest point / long diameter of the posterior papillary muscle at its thickest point - short diameter of the anterior papillary muscle at its thickest point / short diameter of the posterior papillary muscle at its thickest point. If the calculated value approaches 0, it indicates good symmetry between the anterior and posterior papillary muscles. The greater the deviation from 0 (positive or negative), the worse the symmetry between the anterior and posterior papillary muscles, indicating significant structural differences, indicating uneven stress on the mitral valve leaflets and a higher risk of leaflet regurgitation.

[0095] In some optional specific implementations, the target information further includes at least one of the following:

[0096] mitral valve leaflet contracture length;

[0097] residual chordae length;

[0098] Papillary muscle fusion information;

[0099] The major and minor diameters of the annulus.

[0100] Specifically, regarding the length of mitral valve leaflet contracture, when measuring the mitral valve leaflet fiber thickening information, the A1-P1 section, A2-P2 section, and A3-P3 section are also measured. Figure 9 As shown, the image was enlarged and the measuring ruler was retrieved to trace the distance from the anterior and posterior leaflet roots to the leaflet tips, which were recorded as the leaflet lengths A1, A2, A3, P1, P2, and P3 to quantify the leaflet contracture lengths of the characteristic rheumatic mitral valve lesions.

[0101] Regarding the length of the residual chordae tendineae, the layers can be adjusted layer by layer in the three-chamber heart and left ventricular outflow tract sections, such as Figure 10 As shown, zoom in on the image, access the measuring scale, and trace the length between the origin of the anterior and posterior papillary muscles and the leaflets. This is the residual chordae tendineae length of the mitral valve. Similarly, measure the shortest and longest chordae tendineae connecting the anterior and posterior papillary muscles to the leaflets to quantify the extent of chordae tendineae shortening. Excessively short chordae tendineae suggest that surgical repair may require severing excessively restrictive chordae or partially incising the papillary muscles to reconstruct the chordae tendineae.

[0102] Regarding papillary muscle fusion information, the four-chamber heart, coronal left ventricular outflow tract section can be adjusted layer by layer to assess whether the anterior and posterior papillary muscles are clustered or fused into a single thick papillary muscle (e.g. Figure 11 If the papillary muscle is fused into a single papillary muscle, it indicates that the papillary muscle needs to be appropriately incised and released during surgical repair to restore its physiological and anatomical characteristics.

[0103] Regarding the long and short diameters of the valve ring, the complete shape of the mitral valve ring can be obtained by reconstructing the three-chamber heart view, finding the root of the mitral valve leaflet, and reconstructing the mitral valve short axis view at the root level. Figure 12 As shown, the image is magnified to access the measuring scale and measure the mitral valve annulus length and short diameter to obtain the annular length and short diameter data. The annular length and short diameter data can indicate the size of the surgical annulus ring or replacement valve.

[0104] In summary, the embodiments of the present invention can reconstruct and locate the mitral valve lesions before surgery and analyze them through multi-planar reconstruction technology, provide surgical treatment prompts and guidance, and then select the best surgical plan based on the lesion condition.

[0105] Among them, cardiac computed tomography images, also known as cardiac CT images, can clearly display structures such as the mitral valve leaflets, commissures, annuli, chordae tendineae and papillary muscles, and have excellent evaluation effects on the location and degree of calcification. However, echocardiography has defects such as limited acoustic windows and high dependence on the examiner. It has great limitations in evaluating mitral valve calcification and fusion of chordae tendineae and papillary muscles, and these factors significantly affect the successful repair of the mitral valve. Therefore, the embodiment of the present invention can accurately evaluate the degree of mitral valve lesions in RMD patients before surgery by using cardiac CT images, which makes up for the current shortcomings of echocardiography in evaluating mitral valve calcification and subvalvular chordae tendineae and papillary muscle lesions.

[0106] Cardiac CT images can specifically be images of the mitral valve complex (including 3D reconstructed images) acquired through full cardiac cycle axial cardiac CT scans, with a slice thickness of ≤0.75 mm. The scanning range extends from the tracheal carina to the diaphragmatic surface of the heart, 1–2 cm lateral to the cardiac margin on both the left and right sides, encompassing the entire heart. Retrospective ECG-gated scanning is used. The tube voltage is 100 kV, and the tube current is automatically calculated and optimized by Care 4D software. Collimation is 64 × 0.6 mm, slice thickness is 0.75 mm, and interslice spacing is 0.5 mm. Reconstruction convolution kernel is I26f medium smooth ASA, reconstruction window width is 800–1000, window level is 200–300, and tube rotation speed is 0.28 seconds / revolution. Scanning direction is cranio-pedal. Exposure is performed using ECG milliampere modulation, with the high-dose exposure region covering 0%–100% of the RR interval, i.e., the entire RR interval. After the scan is completed, the optimal systolic and diastolic phases are automatically determined based on coronary artery motion to obtain coronary artery information. Manual reconstruction using ECG editing is required when coronary artery visualization is unsatisfactory. Multiphase data from the entire cardiac cycle is reconstructed at a 10% RR interval to visualize aortic and mitral valve motion. The reconstructed area is appropriately narrowed around the valve to reduce redundant data.

[0107] In addition, the non-ionic contrast agent iopromide (iodine concentration 370 mg / ml; Bayer ScheringPharma) was used as the contrast agent. During the examination, the region of interest (ROI) was selected at the aortic root level using bolus tracking. The CT value of the region of interest (ROI) was monitored. When the CT value within the ROI reached 100HU, a scan was automatically triggered with a 6-second delay. A dual-tube, dual-flow, three-phase injection protocol was used. The first phase consisted of a total volume of 60-80 ml of contrast agent, adjusted according to BMI, at a flow rate of 4-5 ml / s. The second phase consisted of a mixture of contrast agent and saline at a ratio of 3:7, totaling 30 ml, with the same injection rate as the first phase. The third phase consisted of a total volume of 30 ml of saline, with the same injection rate as the first phase.

[0108] Step S102: input the target information into a pre-built and trained surgery success probability prediction model for prediction; the surgery success probability prediction model is used to predict the probability of successful mitral valve repair.

[0109] The probability of successful mitral valve repair specifically refers to the probability of good mid- to long-term efficacy after valve repair, or it can also refer to the good effect of mitral valve repair at discharge (transthoracic echocardiography). Figure 2 Cusp area ≥ 1.5 cm 2 , mitral transvalvular gradient ≤ 5 mmHg, mitral regurgitation ≤ 2 / 4) probability.

[0110] Step S103: Obtain the surgical success probability output by the surgical success probability prediction model. This surgical success probability is used to guide the selection of a mitral valve surgical strategy for the target patient. For example, if the surgical success probability output by the model is greater than 0.85, mitral valve repair surgery may be recommended, while if the surgical success probability is less than 0.30, mitral valve replacement surgery may be prioritized.

[0111] The method for predicting the success probability of rheumatic mitral valve surgery provided in this embodiment can locate and determine the condition of mitral valve lesions before surgery, thereby accurately assessing the degree of mitral valve lesions in RMD patients (i.e., target subjects) and accurately predicting the probability of good repair of the mitral valve in RMD patients, thereby providing surgical treatment prompts and guidance, and facilitating personalized selection of the best surgical plan based on the condition of the lesion.

[0112] The embodiments of the present invention propose for the first time specific evaluation indicators: penetrating calcification, papillary muscle symmetry, residual chordae tendineae length, etc. Statistical results show that these factors significantly affect the postoperative effect of mitral valve repair and can be used as standard indicators for subsequent evaluation of mitral valve lesions.

[0113] In summary, the embodiment of the present invention proposes a standardized scheme for evaluating rheumatic mitral valve lesions using cardiac CT, which provides great convenience and objective reference for objective, accurate, and comprehensive preoperative evaluation and quantification of mitral valve lesions and personalized design of surgical plans. Moreover, the standardized scheme for evaluating rheumatic mitral valve lesions using cardiac CT proposed in the embodiment of the present invention has considerable potential for expanding indications. Subsequently, personalized indicators can be integrated on the basis of this scheme for different types of mitral valve diseases (such as degenerative mitral valve lesions) to meet the future needs for the diagnosis and treatment of various types of mitral valve lesions, especially the development of mitral valve interventional diagnosis and treatment.

[0114] In addition, the method for predicting the success probability of rheumatic mitral valve surgery provided by the embodiment of the present invention can form a corresponding software program, and then be integrated into the CT post-processing software together with the artificial intelligence model used, or an independent artificial intelligence module can be developed.

[0115] In the above embodiment, each specific target information of the target object mitral valve can be obtained using image processing technology, or can be obtained using artificial intelligence recognition technology such as deep learning models. When using a deep learning model to identify cardiac CT images to obtain each specific target information, different deep learning models can be used to identify cardiac CT images to obtain different target information. However, this method requires the deployment of multiple deep learning models, and the recognition process is relatively complicated. To this end, an embodiment of the present invention proposes a technology for using a multi-task deep learning model to identify each specific target information.

[0116] Specifically, in some optional specific embodiments, acquiring target information of the mitral valve of the target object includes:

[0117] Using a target information recognition model, recognizing a cardiac computed tomography image of the target object;

[0118] determining the target information based on an output of the target information recognition model;

[0119] The target information recognition model is a multi-task recognition model, specifically including the recognition tasks of the various target information described in the above embodiments (Agatston integral of the mitral valve calcification area; calcification volume integral of the mitral valve calcification area; calcification mass integral of the mitral valve calcification area; location information of the mitral valve calcification area; degree of mitral valve leaflet calcification invasion; residual leaflet thickness in the mitral valve calcification area; thickness information of each area of ​​the anterior leaflet; thickness information of each area of ​​the posterior leaflet; length of mitral valve leaflet contracture; residual chordae tendineae length; papillary muscle fusion information; major and minor diameters of the annulus, etc.). In other words, the recognition of each specific target information constitutes a task. The multi-task recognition model comprises a shared feature network and branch output networks for each task. The shared feature network is used to extract feature information of the cardiac CT image, and the branch output networks for each task predict the corresponding specific target information based on the feature information output by the shared feature network.

[0120] Compared with the technical solution of using different deep learning models to predict different target information respectively, the embodiment of the present invention only needs to deploy one multi-task recognition model to realize the recognition of multiple different specific target information. Moreover, the network for extracting the feature information of cardiac CT images in the multi-task recognition model is shared, which can greatly reduce the requirements for computing power.

[0121] Furthermore, given that different target information needs to be acquired from different regions of cardiac CT images, the multi-task recognition model provided by the present invention employs independent spatial attention layers in different branch output networks, generating attention maps that dynamically weight the feature information output by the shared feature network. During the training of this multi-task recognition model, the attention map can be jointly optimized with the task loss to ensure that the attention area is consistent with the task objective.

[0122] The multi-task recognition model can also dynamically allocate the feature information output by the shared feature network to different branch output networks through a gating mechanism. Specifically, a differentiable ROI Selector can be used to dynamically generate a region of interest (ROI) specific to each task. That is, each task independently generates an ROI mask to solve the problem of multi-task region conflicts. Moreover, the output is a probability mask rather than a hard bounding box, allowing the model to learn regional importance weights. Therefore, in an embodiment of the present invention, each branch output network in the multi-task recognition model includes an independent lightweight convolutional layer, which outputs an ROI probability mask, which is used to perform element-by-element multiplication with the feature information output by the shared feature network to obtain a weighted feature specific to the branch output network, and then the branch output network identifies the corresponding target information based on the exclusive weighted feature.

[0123] In the embodiment of the present invention, the image regions that should be focused on for different tasks are automatically learned through a neural network, without the need to manually preset ROI coordinates or rely on traditional region proposal networks (such as RPN).

[0124] In some optional specific embodiments, the surgery success probability prediction model is an uncertainty-aware probability model, which includes an input layer, a feature extractor, an uncertainty estimation module, and an output layer;

[0125] Among them, the input layer is used to receive the target information; the feature extractor is a Transformer network, which is used to extract features from the target information to obtain corresponding feature information; the uncertainty estimation module includes a Bayesian neural network, which is used to output an uncertain estimation result based on the feature information; the output layer is used to generate a prediction result of the probability of success of the operation based on the feature information and the uncertain estimation result, and output the predicted probability of success of the operation and the uncertain estimation result.

[0126] Specifically, in an embodiment of the present invention, Bayesian neural networks (BNNs) are used in the uncertainty estimation module to perform probabilistic modeling of weights to predict uncertainty. The Bayesian neural network assigns a prior distribution (such as a Gaussian distribution) to the weights and biases of the neural network, and calculates the posterior distribution through Bayesian inference. Since the posterior distribution is usually difficult to solve analytically, variational inference or Markov Chain Monte Carlo Method (MCMC) method is often used for approximation. In other embodiments, multiple models can also be integrated (such as Dropout integration, deep integration) to estimate the uncertainty of the prediction.

[0127] In the embodiment of the present invention, an uncertainty-aware probability model is used when predicting the probability of surgical success, thereby not only predicting the probability of surgical success, but also quantifying the confidence (or uncertainty) of the prediction. In this way, doctors can not only know the probability that RMD patients who choose valve repair surgery have a good prognosis, but also know the degree of grasp of the model on the predicted probability, which is beneficial for doctors to choose appropriate surgical strategies for RMD patients.

[0128] External validation was performed using a cohort of 109 patients undergoing rheumatic mitral valve surgery at Zhongshan People's Hospital. The model demonstrated good predictive performance, with an area under the curve (AUC) of 0.83 (95% CI: 0.72-0.93) (CI stands for confidence interval).

[0129] In other optional embodiments, the surgical success probability prediction model can also be a multimodal graph neural network (Graph Attention Network (GAT) + Cross-Modal Fusion). Alternatively, the surgical success probability prediction model can be obtained by searching for the optimal network topology through algorithms such as Efficient Neural Architecture Search (ENAS) and Differentiable Architecture Search (DARTS).

[0130] In some optional specific embodiments, inputting the target information into a pre-built and trained surgery success probability prediction model for prediction includes:

[0131] obtaining coronary artery disease information and / or pulmonary artery pressure information of the target subject;

[0132] The target information, as well as the coronary artery disease information and / or the pulmonary artery pressure information, are input into the surgery success probability prediction model for prediction.

[0133] The embodiments of the present invention not only evaluate the mitral valve lesions based on cardiac CT, but also combine clinical factors (i.e., including coronary artery disease information and / or pulmonary artery pressure information, etc.) to predict early good mitral valve repair in RMD patients, thereby improving the prediction accuracy and providing systematic guidance for the selection of surgical strategies for RMD patients.

[0134] This embodiment also provides a device for predicting the success probability of rheumatic mitral valve surgery. The device is used to implement the above-mentioned embodiments and preferred embodiments, and the details already described will not be repeated. As used below, the term "module" can refer to a combination of software and / or hardware that implements a predetermined function. Although the devices described in the following embodiments are preferably implemented in software, implementation using hardware, or a combination of software and hardware, is also possible and contemplated.

[0135] This embodiment provides a device for predicting the success probability of rheumatic mitral valve surgery. Figure 13 As shown, including:

[0136] a target information acquisition module 1301 configured to acquire target information of the mitral valve of a target subject, the target information including at least mitral valve calcification information, mitral valve leaflet fiber thickening information, and papillary muscle symmetry information; the target information is measured based on a reconstructed three-dimensional image of the mitral valve, the three-dimensional image of the mitral valve being reconstructed based on a cardiac computed tomography image of the target subject;

[0137] Prediction module 1302 is used to input the target information into a pre-built and trained surgery success probability prediction model for prediction; the surgery success probability prediction model is used to predict the probability of successful mitral valve repair;

[0138] The surgery success probability acquisition module 1303 is used to obtain the surgery success probability output by the surgery success probability prediction model, and the surgery success probability is used to guide the selection of the mitral valve surgery strategy for the target subject.

[0139] In some optional embodiments, the mitral valve calcification information includes at least one of the following:

[0140] Agatston score of mitral valve calcification area;

[0141] calcification volume integral of the mitral valve calcification area;

[0142] Calcification mass integral of the mitral valve calcification area;

[0143] Location information of mitral valve calcification areas;

[0144] The degree of mitral valve leaflet calcification invasion;

[0145] Thickness of residual leaflets in the mitral valve calcification area.

[0146] In some optional embodiments, the device for predicting the probability of success of rheumatic mitral valve surgery further comprises:

[0147] The leaflet rupture probability determination module is used to determine the probability of mitral valve leaflet rupture when surgical repair surgery is performed to remove the calcification foci on the leaflets based on the degree of mitral valve leaflet calcification invasion and / or the residual leaflet thickness at the mitral valve calcification site.

[0148] In some optional embodiments, the mitral valve leaflet fiber thickening information includes at least one of the following: thickness information of the A1 area of ​​the anterior leaflet; thickness information of the A2 area of ​​the anterior leaflet; thickness information of the A3 area of ​​the anterior leaflet; thickness information of the P1 area of ​​the posterior leaflet; thickness information of the P2 area of ​​the posterior leaflet; thickness information of the P3 area of ​​the posterior leaflet;

[0149] The thickness information of the A1 region of the anterior lobe includes: thickness information of the zona pellucida corresponding to the A1 region of the anterior lobe, thickness information of the zona sclera corresponding to the A1 region of the anterior lobe, and overall thickness information of the A1 region of the anterior lobe;

[0150] The thickness information of the A2 area of ​​the anterior leaflet includes: the thickness information of the zona pellucida corresponding to the A2 area of ​​the anterior leaflet, the thickness information of the zona crassa corresponding to the A2 area of ​​the anterior leaflet, and the overall thickness information of the A2 area of ​​the anterior leaflet;

[0151] The thickness information of the A3 area of ​​the anterior leaflet includes: the thickness information of the zona pellucida corresponding to the A3 area of ​​the anterior leaflet, the thickness information of the zona crassa corresponding to the A3 area of ​​the anterior leaflet, and the overall thickness information of the A3 area of ​​the anterior leaflet;

[0152] The thickness information of the P1 region of the posterior leaflet includes: the thickness information of the rough zone portion corresponding to the P1 region of the posterior leaflet and the overall thickness information of the P1 region of the posterior leaflet;

[0153] The thickness information of the P2 region of the posterior leaflet includes: thickness information of the rough zone portion corresponding to the P2 region of the posterior leaflet and overall thickness information of the P2 region of the posterior leaflet;

[0154] The thickness information of the P3 region of the posterior lobe includes: thickness information of the rough zone portion corresponding to the P3 region of the posterior lobe and overall thickness information of the P3 region of the posterior lobe.

[0155] In some optional implementations, the target information further includes at least one of the following:

[0156] mitral valve leaflet contracture length;

[0157] residual chordae length;

[0158] Papillary muscle fusion information;

[0159] The major and minor diameters of the annulus.

[0160] In some optional implementations, the target information acquisition module includes:

[0161] an identification unit, configured to identify the cardiac computed tomography image of the target object using a target information identification model;

[0162] a target information determining unit, configured to determine the target information based on an output of the target information recognition model;

[0163] Among them, the target information recognition model is a multi-task recognition model, which has a shared feature network and multiple branch output networks. The shared feature network is used to extract the feature information of the cardiac computed tomography image, and the multiple branch output networks respectively predict different target information based on the feature information output by the shared feature network.

[0164] In some optional embodiments, the surgery success probability prediction model is an uncertainty-aware probability model, which includes an input layer, a feature extractor, an uncertainty estimation module, and an output layer;

[0165] Among them, the input layer is used to receive the target information; the feature extractor is a Transformer network, which is used to extract features from the target information to obtain corresponding feature information; the uncertainty estimation module includes a Bayesian neural network, which is used to output an uncertain estimation result based on the feature information; the output layer is used to generate a prediction result of the probability of success of the operation based on the feature information and the uncertain estimation result, and output the predicted probability of success of the operation and the uncertain estimation result.

[0166] In some optional embodiments, the prediction module includes:

[0167] a clinical factor acquisition unit, configured to acquire coronary artery disease information and / or pulmonary artery pressure information of the target subject;

[0168] The comprehensive prediction unit is used to input the target information, the coronary artery disease information and / or the pulmonary artery pressure information into the operation success probability prediction model for prediction.

[0169] The further functional description of each of the above modules and units is the same as that of the above corresponding embodiments and will not be repeated here.

[0170] The device for predicting the success probability of rheumatic mitral valve surgery in this embodiment is presented in the form of a functional unit, where the unit refers to an ASIC (Application Specific Integrated Circuit) circuit, a processor and memory that executes one or more software or fixed programs, and / or other devices that can provide the above functions.

[0171] The embodiment of the present invention also provides a medical device having the above Figure 13 The device shown is used to predict the probability of success of rheumatic mitral valve surgery.

[0172] See also Figure 14 , Figure 14 is a structural diagram of a medical device provided by an optional embodiment of the present invention, such as Figure 14 As shown, the medical device includes: one or more processors 10, a memory 20, and interfaces for connecting various components, including high-speed interfaces and low-speed interfaces. The various components are connected to each other using different buses and can be installed on a common motherboard or installed in other ways as needed. The processor can process instructions executed in the medical device, including instructions stored in or on the memory to display graphical information of the GUI on an external input / output device (such as a display device coupled to the interface). In some optional embodiments, if necessary, multiple processors and / or multiple buses can be used together with multiple memories and multiple memories. Figure 14 A processor 10 is taken as an example.

[0173] The processor 10 may be a central processing unit, a network processor, or a combination thereof. The processor 10 may further include a hardware chip. The hardware chip may be an application-specific integrated circuit, a programmable logic device, or a combination thereof. The programmable logic device may be a complex programmable logic device, a field programmable gate array, a general purpose array logic, or any combination thereof.

[0174] The memory 20 stores instructions that can be executed by at least one processor 10, so that the at least one processor 10 executes the method shown in the above embodiment.

[0175] The memory 20 may include a program storage area and a data storage area, wherein the program storage area may store an operating system and application programs required for at least one function; the data storage area may store data created based on the use of the medical device, etc. In addition, the memory 20 may include a high-speed random access memory and may also include a non-transient memory, such as at least one disk storage device, a flash memory device, or other non-transient solid-state storage device. In some optional embodiments, the memory 20 may optionally include a memory remotely located relative to the processor 10, and these remote memories may be connected to the medical device via a network. Examples of the aforementioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0176] The memory 20 may include a volatile memory, such as a random access memory; the memory may also include a non-volatile memory, such as a flash memory, a hard disk or a solid-state drive; the memory 20 may also include a combination of the above types of memory.

[0177] The medical device further includes an input device 30 and an output device 40. The processor 10, the memory 20, the input device 30 and the output device 40 may be connected via a bus or other means. Figure 14 The bus connection is taken as an example.

[0178] The input device 30 can receive input digital or character information and generate key signal input related to user settings and function control of the medical device, such as a touch screen, a keypad, a mouse, a trackpad, a touch pad, an indicator stick, one or more mouse buttons, a trackball, a joystick, etc. The output device 40 can include a display device, an auxiliary lighting device (e.g., an LED), and a tactile feedback device (e.g., a vibration motor). The above-mentioned display device includes but is not limited to a liquid crystal display, a light emitting diode, a display, and a plasma display. In some optional embodiments, the display device can be a touch screen.

[0179] The medical device also includes a communication interface for the medical device to communicate with other devices or a communication network.

[0180] The embodiment of the present invention also provides a computer-readable storage medium. The above-mentioned method according to the embodiment of the present invention can be implemented in hardware, firmware, or implemented as a computer code that can be recorded in a storage medium, or implemented as a computer code that is originally stored in a remote storage medium or a non-temporary machine-readable storage medium and downloaded through a network and will be stored in a local storage medium, so that the method described herein can be stored in such software processing on a storage medium using a general-purpose computer, a dedicated processor, or programmable or dedicated hardware. Among them, the storage medium can be a magnetic disk, an optical disk, a read-only storage memory, a random access memory, a flash memory, a hard disk or a solid-state drive, etc.; further, the storage medium can also include a combination of the above-mentioned types of memory. It can be understood that a computer, a processor, a microprocessor controller or programmable hardware includes a storage component that can store or receive software or computer code. When the software or computer code is accessed and executed by a computer, a processor or hardware, the method shown in the above embodiment is implemented.

[0181] A portion of the present invention may be applied as a computer program product, such as a computer program instruction, which, when executed by a computer, can call or provide the method and / or technical solution according to the present invention through the operation of the computer. Those skilled in the art should understand that the form in which the computer program instruction exists in a computer-readable medium includes, but is not limited to, a source file, an executable file, an installation package file, etc. Accordingly, the way in which the computer program instruction is executed by the computer includes, but is not limited to: the computer directly executes the instruction, or the computer compiles the instruction and then executes the corresponding compiled program, or the computer reads and executes the instruction, or the computer reads and installs the instruction and then executes the corresponding installed program. Here, the computer-readable medium may be any available computer-readable storage medium or communication medium that can be accessed by the computer.

[0182] Although the embodiments of the present invention have been described with reference to the accompanying drawings, those skilled in the art may make various modifications and variations without departing from the spirit and scope of the present invention. Such modifications and variations are all within the scope defined by the appended claims.

Claims

1. A method for predicting the success probability of rheumatic mitral valve surgery, characterized in that: The method comprises: acquiring target information of a mitral valve of a target subject, the target information including at least mitral valve calcification information, mitral valve leaflet fiber thickening information, and papillary muscle symmetry information; the target information is measured based on a reconstructed three-dimensional image of the mitral valve, the three-dimensional image of the mitral valve being reconstructed based on a cardiac computed tomography image of the target subject; Inputting the target information into a pre-built and trained surgical success probability prediction model for prediction; the surgical success probability prediction model is used to predict the probability of successful mitral valve repair; The surgical success probability output by the surgical success probability prediction model is obtained, and the surgical success probability is used to guide the selection of a mitral valve surgical strategy for the target subject.

2. The method according to claim 1, characterized in that The mitral valve calcification information includes at least one of the following: Agatston score of mitral valve calcification area; calcification volume integral of the mitral valve calcification area; Calcification mass integral of the mitral valve calcification area; Location information of mitral valve calcification areas; The degree of mitral valve leaflet calcification invasion; Thickness of residual leaflets in the mitral valve calcification area.

3. The method according to claim 2, characterized in that Also includes: Based on the degree of mitral valve leaflet calcification and / or the thickness of the residual leaflet at the mitral valve calcification site, the probability of mitral valve leaflet rupture when surgical repair is performed to remove the calcification on the leaflet is determined.

4. The method according to claim 1, wherein The mitral valve leaflet fiber thickening information includes at least one of the following: thickness information of the A1 area of ​​the anterior leaflet; thickness information of the A2 area of ​​the anterior leaflet; thickness information of the A3 area of ​​the anterior leaflet; thickness information of the P1 area of ​​the posterior leaflet; thickness information of the P2 area of ​​the posterior leaflet; thickness information of the P3 area of ​​the posterior leaflet; The thickness information of the A1 region of the anterior lobe includes: thickness information of the zona pellucida corresponding to the A1 region of the anterior lobe, thickness information of the zona sclera corresponding to the A1 region of the anterior lobe, and overall thickness information of the A1 region of the anterior lobe; The thickness information of the A2 area of ​​the anterior leaflet includes: the thickness information of the zona pellucida corresponding to the A2 area of ​​the anterior leaflet, the thickness information of the zona crassa corresponding to the A2 area of ​​the anterior leaflet, and the overall thickness information of the A2 area of ​​the anterior leaflet; The thickness information of the A3 area of ​​the anterior leaflet includes: the thickness information of the zona pellucida corresponding to the A3 area of ​​the anterior leaflet, the thickness information of the zona crassa corresponding to the A3 area of ​​the anterior leaflet, and the overall thickness information of the A3 area of ​​the anterior leaflet; The thickness information of the P1 region of the posterior leaflet includes: the thickness information of the rough zone portion corresponding to the P1 region of the posterior leaflet and the overall thickness information of the P1 region of the posterior leaflet; The thickness information of the P2 region of the posterior leaflet includes: thickness information of the rough zone portion corresponding to the P2 region of the posterior leaflet and overall thickness information of the P2 region of the posterior leaflet; The thickness information of the P3 region of the posterior lobe includes: thickness information of the rough zone portion corresponding to the P3 region of the posterior lobe and overall thickness information of the P3 region of the posterior lobe.

5. The method according to claim 1, characterized in that The target information also includes at least one of the following: mitral valve leaflet contracture length; residual chordae length; Papillary muscle fusion information; The major and minor diameters of the annulus.

6. The method according to claim 1, characterized in that The acquiring target information of the mitral valve of the target object includes: Using a target information recognition model, recognizing a cardiac computed tomography image of the target object; determining the target information based on an output of the target information recognition model; Among them, the target information recognition model is a multi-task recognition model, which has a shared feature network and multiple branch output networks. The shared feature network is used to extract the feature information of the cardiac computed tomography image, and the multiple branch output networks respectively predict different target information based on the feature information output by the shared feature network.

7. The method according to claim 1, characterized in that The surgery success probability prediction model is an uncertainty-aware probability model, which includes an input layer, a feature extractor, an uncertainty estimation module, and an output layer; Among them, the input layer is used to receive the target information; the feature extractor is a Transformer network, which is used to extract features from the target information to obtain corresponding feature information; the uncertainty estimation module includes a Bayesian neural network, which is used to output an uncertain estimation result based on the feature information; the output layer is used to generate a prediction result of the probability of success of the operation based on the feature information and the uncertain estimation result, and output the predicted probability of success of the operation and the uncertain estimation result.

8. The method according to claim 1, characterized in that Inputting the target information into a pre-built and trained surgery success probability prediction model for prediction includes: obtaining coronary artery disease information and / or pulmonary artery pressure information of the target subject; The target information, as well as the coronary artery disease information and / or the pulmonary artery pressure information, are input into the surgery success probability prediction model for prediction.

9. A device for predicting the success probability of rheumatic mitral valve surgery, characterized in that: The device comprises: a target information acquisition module, configured to acquire target information of the mitral valve of a target subject, the target information including at least mitral valve calcification information, mitral valve leaflet fiber thickening information, and papillary muscle symmetry information; the target information is measured based on a reconstructed three-dimensional image of the mitral valve, the three-dimensional image of the mitral valve being reconstructed based on a cardiac computed tomography image of the target subject; A prediction module is used to input the target information into a pre-built and trained surgery success probability prediction model for prediction; the surgery success probability prediction model is used to predict the probability of successful mitral valve repair; The operation success probability acquisition module is used to obtain the operation success probability output by the operation success probability prediction model, and the operation success probability is used to guide the selection of the mitral valve surgery strategy for the target object.

10. A medical device, characterized in that: include: A memory and a processor, wherein the memory and the processor are communicatively connected to each other, the memory stores computer instructions, and the processor executes the method for predicting the success probability of rheumatic mitral valve surgery according to any one of claims 1 to 8 by executing the computer instructions.

Images