Multi-modal clinical data placenta implantation evaluation system

Abstract

The placenta implantation sign evaluation device comprises a collection unit, a screening unit, a fusion unit, a placenta implantation sign evaluation unit and a placenta implantation sign evaluation unit. The collection unit collects and preprocesses multiple blood flow indexes and inputs the multiple blood flow indexes into the screening unit. The screening unit is used for extracting effective blood flow indexes for placenta implantation sign typing from the plurality of blood flow indexes by utilizing a first mode and / or a second mode, and inputting the effective blood flow indexes into the fusion unit; and the fusion unit inputs the effective blood flow index and the clinical indexes into respective evaluators, and fuses the placenta implantation sign typing results output by the evaluators to obtain a placenta implantation sign evaluation result. By means of the placenta implantation sign evaluation device, a plurality of blood flow indexes for accurately evaluating placenta implantation ultrasonic signs can be obtained, and the blood flow indexes and clinical information are fused to evaluate placenta implantation development conditions of a subject in different pregnancy periods.

Description

Technical Field

[0001] The present application belongs to the technical field of placenta accreta detection, and specifically relates to a multimodal clinical data placenta accreta assessment system. Background Art

[0002] Placenta accreta is a pathological condition in which the placental villi abnormally invade the myometrium and even penetrate the serosa. This can lead to fatal postpartum hemorrhage and infection. To control the bleeding, the pregnancy must be terminated early, which increases the risk of complications in premature babies. In addition, placenta accreta can lead to insufficient blood supply to the placenta, limiting normal fetal growth. Placenta accreta and its associated symptoms during pregnancy can cause changes in blood flow in the uterus and placenta of pregnant women. For example, the number of endothelial cells in the peripheral blood and umbilical cord blood of patients with preeclampsia decreases and their function is impaired, uterine artery hemodynamic parameters are significantly higher than those of normal pregnant women, and CN102058435B uses blood flow characteristics such as peripheral vascular resistance and pulse waves to predict gestational hypertension.

[0003] In the past, due to limitations in detection methods, the detection rate of indicators such as uterine blood flow was low. With the development of ultrasound imaging technology, technologies such as color Doppler energy imaging and three-dimensional ultrasound have been widely used for endometrial blood flow detection. The advent of three-dimensional ultrasound has further effectively compensated for the shortcomings of two-dimensional ultrasound. It can synthesize and analyze image information of the placenta's coronal surface through computer technology to provide three-dimensional stereoscopic images, thereby effectively measuring endometrial volumetric blood flow parameters such as the vascular index (VI), blood flow index (FI), and vascular flow index (VFI). These blood flow parameters are difficult to obtain using traditional detection methods. These blood flow parameters have the potential to analyze the development of placental implantation.

[0004] However, there is still a lack of methods to analyze or detect placenta accreta using hemodynamic parameters. Currently, the diagnosis of accreta, accreta, and percreta accreta is still mainly based on manual or intelligent analysis of ultrasound images, which requires more manpower or computing resources, has relatively high requirements for image data, and the stability of the assessment may be difficult to guarantee.

[0005] Therefore, there is a need for an intelligent system that uses hemodynamic parameter assessment to classify placenta accreta. Summary of the Invention

[0006] In order to solve the problems existing in the prior art, the purpose of this application is to provide a placenta accreta sign assessment device, which can use an acquisition unit to collect multiple blood flow indicators of the placenta from ultrasound images, use a screening unit to extract effective blood flow indicators that can be used to effectively distinguish people with different types of placenta accreta and healthy people from the blood flow indicators of multiple different blood vessels at different time periods of pregnancy, and combine clinical parameters of other modalities to predict multiple signs of placenta accreta to obtain an assessment result of the placenta accreta sign.

[0007] Specifically, this application involves the following aspects:

[0008] According to one aspect of the present application, a placenta accreta sign assessment device is provided, comprising: an acquisition unit, configured to acquire and preprocess a plurality of blood flow indicators and a plurality of clinical indicators, input the preprocessed plurality of blood flow indicators into a screening unit, and input the plurality of clinical indicators into a fusion unit; a screening unit, configured to extract effective blood flow indicators for placenta accreta sign classification from the plurality of blood flow indicators using a first mode and / or a second mode, and input the effective blood flow indicators into the fusion unit; and a fusion unit, configured to input the effective blood flow indicators and the plurality of clinical indicators into respective evaluators, and fuse the placenta accreta sign classification results output by the evaluators to obtain a placenta accreta sign assessment result.

[0009] According to some embodiments of the present application, the first mode includes: inputting multiple blood flow indicators into a classifier, in response to the evaluation index value of the classifier not reaching a preset standard, adjusting the number and / or type of blood flow indicators input to the classifier, and in response to the evaluation index value of the classifier reaching a preset standard, setting the blood flow indicator currently input to the classifier as a valid blood flow indicator.

[0010] According to some embodiments of the present application, the first mode includes: inputting multiple blood flow indicators into a classifier, sorting the placenta implantation sign classification weight values ​​of the multiple blood flow indicators from low to high; eliminating blood flow indicators in order, retraining the classifier and sorting the multiple blood flow indicators from low to high based on the new weight values, in response to an increase in the classifier evaluation index value, continuing the classifier training, weight value sorting and sequential elimination of blood flow indicators until the classifier evaluation index value no longer increases, in response to a decrease in the classifier evaluation index value, reintroducing the eliminated blood flow indicator and eliminating the next blood flow indicator in order, and continuing the classifier training, weight value sorting and evaluation index value comparison steps; and taking the last retained blood flow indicator as the valid blood flow indicator.

[0011] According to some embodiments of the present application, the second mode includes: performing inter-group analysis of placenta accreta sign typing in multiple time intervals on multiple blood flow indicators to determine that the blood flow indicators among the multiple blood flow indicators have significant differences between patients with placenta accreta sign and healthy people are effective blood flow indicators.

[0012] According to some embodiments of the present application, fusing the placenta accreta sign classification results output by the evaluator to obtain a placenta accreta sign evaluation result includes: inputting multiple effective blood flow indicators and multiple clinical indicators into a first evaluator and a second evaluator, respectively, to obtain a first classification result and a second classification result, respectively; assigning a first weight and a second weight to the first classification result and the second classification result according to the evaluation index values ​​of the first evaluator and the second evaluator, respectively, to weight the first classification result and the second classification result to obtain a placenta accreta sign evaluation result.

[0013] According to some embodiments of the present application, fusing the placenta accreta sign classification results output by the evaluator to obtain the placenta accreta sign classification results includes: inputting multiple effective blood flow indicators and multiple clinical indicators into a first evaluator and a second evaluator, respectively, to obtain a first classification result and a second classification result, respectively; inputting the first classification result and the second classification result into a third evaluator, training the third evaluator so that its evaluation index values ​​reach preset standards, and outputting the third classification result from the third evaluator as the placenta accreta sign evaluation result.

[0014] According to some embodiments of the present application, signs of placenta accreta include: placental position, placental thickness, retroplacental hypoechoic zone, bladder line, placental crypt, placental base blood flow, cervical sinusoids and / or cervical morphology.

[0015] According to some embodiments of the present application, multiple blood flow indicators include: vascular index, blood flow index, vascular blood flow index, resistance index, pulsatility index and / or the ratio of end-systolic peak to end-diastolic peak, which are collected separately in 7 time intervals of 11-16 weeks, 16-20 weeks, 20-24 weeks, 24-28 weeks, 28-32 weeks, 32-36 weeks and 38 weeks and above during pregnancy.

[0016] According to some embodiments of the present application, the resistance index, pulsatility index and end-systolic peak to end-diastolic peak ratio include the resistance index, pulsatility index and end-systolic peak to end-diastolic peak ratio of the umbilical artery, the right uterine artery and the left uterine artery respectively.

[0017] According to another aspect of the present application, a multimodal placenta accreta assessment system is provided, comprising: a sign prediction unit, which collects multiple clinical indicators and multiple blood flow indicators of a subject and fuses the multiple clinical indicators and multiple blood flow indicators, predicts the classification of each placenta accreta sign among the subject's multiple placenta accreta signs based on the fused indicators, and inputs the prediction result into the evaluation unit; and an evaluation unit, which assigns a respective third weight to each of the multiple placenta accreta signs, receives the prediction result, and weightedly fuses the multiple placenta accreta signs to obtain a placenta accreta assessment result for the subject.

[0018] According to some embodiments of the present application, multiple blood flow indicators include: the left uterine artery pulsatility index and the ratio of the left uterine artery end-systolic peak to the end-diastolic peak at 11-16 weeks of pregnancy; the ratio of the umbilical artery end-systolic peak to the end-diastolic peak at 16-20 weeks of pregnancy; the left uterine artery pulsatility index and the ratio of the left uterine artery end-systolic peak to the end-diastolic peak at 24-28 weeks of pregnancy; the change in the umbilical artery resistance index between 16 and 38 weeks of pregnancy; the change in the ratio of the left uterine artery end-systolic peak to the end-diastolic peak between 11 and 32 weeks of pregnancy; and the change in the endometrial volume vascular index and vascular blood flow index between 11 and 38 weeks of pregnancy.

[0019] According to some embodiments of the present application, the multiple clinical indicators include: age, gravidity, medical history information and / or gestational age at delivery.

[0020] According to some embodiments of the present application, the sign prediction unit: utilizes the collection unit of the aforementioned placenta accreta sign evaluation device to collect multiple clinical indicators and multiple blood flow indicators of the subject; and utilizes the fusion unit of the aforementioned placenta accreta sign evaluation device to fuse the multiple clinical indicators and multiple blood flow indicators, and uses an evaluator to predict the typing of each placenta accreta sign among the multiple placenta accreta signs based on the fused indicators.

[0021] The placenta accreta sign assessment device provided in the present application can be used to extract sensitive indicators of placenta accreta signs from multiple blood flow indicators of patients with known placenta accreta results, and the extracted blood flow indicators and other clinical examination results can be used to train an evaluator for placenta accreta signs. After obtaining effective blood flow indicators and evaluators, the multimodal placenta accreta assessment system provided in the present application can predict the classification of multiple placenta accreta signs based on input data provided by a new subject, and calculate the subject's placenta accreta result based on the importance of each placenta accreta sign to the placenta accreta severity assessment. Therefore, the placenta accreta assessment result can be automatically obtained without in-depth analysis of ultrasound images, providing a stable and rapid early screening and evaluation plan for placenta accreta from a new perspective. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] Figure 1 The figure shows a schematic structural diagram of a placenta accreta sign assessment device according to an embodiment of the present application.

[0023] Figure 2 The figure shows a flow chart of screening effective blood flow indicators according to an embodiment of the present application.

[0024] Figure 3 A first schematic diagram of blood flow index values ​​for different groups of people according to an embodiment of the present application is illustrated.

[0025] Figure 4A second schematic diagram of blood flow index values ​​for different groups of people according to an embodiment of the present application is illustrated.

[0026] Figure 5 A third schematic diagram illustrating blood flow index values ​​for different groups of people according to an embodiment of the present application is shown.

[0027] Figure 6 A fourth schematic diagram illustrating blood flow index values ​​for different groups of people according to an embodiment of the present application is shown. DETAILED DESCRIPTION

[0028] The present application is further described below with reference to examples. It should be understood that the examples are only used to further illustrate and explain the present application and are not intended to limit the present application.

[0029] Unless otherwise defined, technical and scientific terms used in this specification have the same meaning as those commonly understood by those skilled in the art. Although methods and materials similar or identical to those described herein may be used in experiments or practical applications, the materials and methods are described herein below. In the event of a conflict, the present specification, including definitions, will prevail. In addition, the materials, methods, and examples are provided for illustrative purposes only and are not intended to be limiting. The present application is further described below with reference to specific examples, which are not intended to limit the scope of this application.

[0030] definition

[0031] Signs of placenta accreta

[0032] The location or structure of placenta implantation that may affect imaging or pathology includes the location of the placenta, the thickness of the placenta, the shape of the retroplacental hypoechoic zone, the shape of the bladder line, the condition of the placental invagination, the shape of the blood flow at the base of the placental junction, the morphology of the cervical sinusoids and cervix, etc. The detection and classification of these signs are important bases for judging the pathological characteristics and development of placenta implantation under non-intraoperative direct visualization. According to the classification of different signs, that is, the severity, the individual's placenta implantation type can be comprehensively evaluated to distinguish whether it is a healthy state or a milder early adhesion type, or a more dangerous implantation type or perforation type.

[0033] Application Overview

[0034] As mentioned above, although many hemodynamic parameters have been used for clinical detection or early screening of diseases during pregnancy, such as gestational hypertension, there is still a lack of methods or systems for interpreting signs of placenta accreta. Specifically, the current ultrasound assessment of placenta accreta includes assessments of blood flow-related items, but they are generally subjective, difficult to maintain consistency and stability, and are only used to assist in the interpretation of ultrasound images. Since the current ultramicro blood flow imaging technology can display the three-dimensional distribution of blood vessels in the placenta accreta area, the placenta accreta range and its blood perfusion in a more intuitive way, and sensitively capture the low-speed perforating vessels in the placenta accreta, thereby more realistically displaying the blood perfusion of the placenta accreta area and accurately distinguishing the boundary between the uterine myometrium and the placenta accreta area, it is completely feasible to use blood flow parameters to evaluate the intelligent assessment of placenta accreta.

[0035] Based on the above considerations, the present application provides a placenta accreta sign assessment device and a multimodal placenta accreta assessment system based on the device. The placenta accreta sign assessment device uses its acquisition unit to collect and preprocess multiple blood flow indicators from ultrasound images and multiple clinical indicators from medical records. The multiple blood flow indicators can reflect the severity of the placenta accreta sign, and the multiple clinical indicators are supplementary information for placenta accreta classification from additional modalities. The screening unit is used to further obtain the multiple blood flow indicators after preprocessing, and the multiple clinical indicators are directly input into the fusion unit. After the screening unit is used to extract the effective classification indicators from the multiple blood flow indicators, the fusion unit uses the easily accessible clinical indicators to supplement the general semantic features that may be lost in the blood flow parameters, thereby improving the accuracy of placenta accreta classification.

[0036] In the screening unit, the present application can use two different screening modes to extract effective blood flow indicators for placenta implantation sign classification from the multiple blood flow indicators. The two modes can be used to screen blood flow indicators that are suitable for some modes respectively and then merge to obtain redundant indicator information to improve the robustness of effective blood flow characteristics. The two modes can also be used to screen the same indicators and then take the intersection to obtain a high-sensitivity blood flow indicator. After the implementation of different forms of schemes, the extracted effective blood flow indicators are input into the fusion unit.

[0037] In the fusion unit, the effective blood flow index and the multiple clinical indicators are input into their respective evaluators. At the decision layer, the placenta accreta sign classification results output by the evaluators are weighted and fused to produce a placenta accreta sign assessment result. This effectively addresses the low stability and potential loss of key features associated with single-modality blood flow data sign assessment. The multimodal placenta accreta assessment system directly utilizes the extracted effective blood flow index and fuses multiple clinical indicators to obtain multiple placenta accreta sign classification results. The weighting of each placenta accreta sign for placenta accreta severity assessment is then set and adjusted, ultimately resulting in a placenta accreta assessment result based on multiple placenta accreta sign classification results. This approach is more objective and stable than traditional manual interpretation of placenta accreta, and offers lower data acquisition, processing, and training costs than image segmentation and classification schemes, thus offering greater practicality.

[0038] After introducing the basic principles of the present application, various non-limiting embodiments of the present application will be described in detail with reference to the accompanying drawings.

[0039] Exemplary Systems

[0040] Figure 1 FIG2 is a block diagram of a placenta accreta sign assessment device according to an embodiment of the present application.

[0041] refer to Figure 1 The placenta accreta sign assessment device provided in this application includes the following parts.

[0042] The acquisition unit 1 is used to acquire and pre-process multiple blood flow indicators and multiple clinical indicators, input the pre-processed multiple blood flow indicators into the screening unit, and input the multiple clinical indicators into the fusion unit.

[0043] The collection unit of the device described in this application has different extraction methods for different types of blood flow indicators. First of all, it should be clarified that the blood flow indicators selected for typing signs of placenta accreta include two categories: volumetric blood flow parameters of the endometrium and specific blood vessel blood flow parameters obtained through blood flow monitoring and analysis. For different categories of blood flow indicators, the collection unit provides different collection schemes:

[0044] Endometrial volumetric blood flow parameters include the aforementioned vascular index (VI), blood flow index (FI), and vascular blood flow index (VFI). VI reflects the density of vascular distribution within the region of interest (ROI) of an ultrasound image, such as the endometrium, and is calculated as the ratio of the color pixel area of ​​the Doppler image to the area of ​​the ROI. FI represents blood flow intensity, and is calculated as the mean Doppler blood flow energy signal. VFI is a weighted fusion of the above two indicators. For clinical energy Doppler ultrasound equipment, the acquisition unit obtains the image captured by the device, performs binarization processing, and integrates the blood flow signal and blood flow energy to obtain the signal area (VI) and energy magnitude (FI). For color Doppler ultrasound images, the color pixel ratio (VI) and mean signal intensity (FI) of the endometrial region can be directly calculated. For contrast-enhanced ultrasound, the area under the curve (VI) of the contrast agent's TIC curve is integrated to obtain the peak intensity (FI). For other common clinical three-dimensional ultrasound systems on the market, their supporting software can mostly directly output these three indicators. This provides some of the blood flow indicators.

[0045] Blood flow parameters obtained by blood flow monitoring and analysis methods may include the Resistance Index (RI), which reflects the proportional relationship between the amplitude of blood flow velocity changes and the end-systolic peak velocity during the cardiac cycle. An increase in peripheral circulatory impedance will accelerate its rate of decline, thus reflecting the relative increase in peripheral impedance due to poor cardiac function or deeper placental implantation. The Pulsatility Index (PI) represents the average of the maximum frequency shift of the velocity waveform (end-systolic peak - end-diastolic peak) within a cardiac cycle. It can reflect the proportional relationship between the amplitude of the decrease in diastolic blood flow velocity and the average velocity, reflecting vascular status from another perspective. The ratio of the end-systolic peak (S) to the end-diastolic peak (D) (S / D) is the amplitude of blood flow velocity changes during the cardiac cycle and can reflect placental blood flow function. It is easy to see that the above three blood flow parameters are all based on the two parameters S and D. In this application, these three parameters are located in the umbilical artery, the left uterine artery, and the right uterine artery, three vascular areas where significant changes will occur during the development of placental implantation. The S, D, and blood flow velocity values ​​throughout the cardiac cycle are obtained by an ultrasonic Doppler probe (or alternatively, arterial puncture) and sent to the acquisition unit for calculation of the corresponding parameters.

[0046] It should be noted that the manifestation of placenta accreta and the degree of placenta accreta reflected by the above-mentioned indicators will also vary in different time periods of pregnancy. If the indicators are not collected by time period and the average value of the entire time period or the indicator value of the early pregnancy is used to classify the signs of placenta accreta, it is likely that good results will not be achieved and the unique manifestations of placenta accreta in different time periods will be ignored. Therefore, the above-mentioned blood flow parameters of this application are collected separately during multiple different gestational time periods. For example, if the individual pregnancy is divided into n time periods for collection, the collection unit will finally collect and calculate 12n blood flow indicators, including VI, FI, VFI, RI, PI and S / D of the umbilical artery, RI, PI and S / D of the right uterine artery, and RI, PI and S / D of the left uterine artery in each of the n time periods. The screening unit will effectively extract the blood flow parameters in each time period, thereby obtaining effective blood flow indicators in the key window period for the classification of signs of placenta accreta.

[0047] For the time period of collecting blood flow index, it is mainly considered that the pregnancy starts 10 weeks later (i.e., the fetal period, the placenta is not yet fully established before 10 weeks). During the 10-15 period, the villi branching increases, the placenta thickens, and the maternal-fetal circulation is established. After 15 weeks, the placental ultrasound mark is clearly manifested. After 20 weeks, the resistance of the uterine artery decreases, and the placental perfusion volume increases. After 28 weeks, the above-mentioned blood flow index can obviously reflect the aging of the placental structure or functional abnormality. Therefore, it is preferred to divide the pregnancy period into multiple time intervals at 5-week intervals to collect blood flow index, so that the placenta in each time interval is different in morphology and function, and ensure that the blood flow index covers different periods of the placenta to reflect the various possible developments of placenta implantation.

[0048] The data extracted by the acquisition unit from the clinical ultrasound device can be in DICOM format, or in a dynamic sequence format or 3D format of DICOM, or in a video format or image format data of Doppler or contrast imaging. By establishing a direct or indirect data transmission interface with the ultrasound device (for example, usually a USB, HDMI or TF, U disk or other data acquisition storage card), the relevant parameters in the DICOM or image frame, such as blood flow signal intensity, energy intensity, S value, D value, etc., are received and calculated to obtain the multiple blood flow indicators required by the screening unit. In addition, the multiple clinical indicators collected by the acquisition unit can include basic indicators with a high correlation with placental diseases, such as the age, gravidity, medical history information and / or gestational age of the individual from whom the blood flow parameters are signed. These indicators are usually easy to obtain, and the acquisition unit can integrate existing technologies (such as the basic network of language models such as BioBERT and ClinicalBERT or the TF-IDF algorithm) to directly extract the word information of these indicators from the individual's medical record report.

[0049] Before screening multiple blood flow indicators, the acquisition unit can also perform various preprocessing operations, including outlier removal, missing value filling, normalization, and labeling, to ensure the data quality of multiple blood flow indicators. Before inputting multiple clinical indicators into the fusion unit, the acquisition unit also needs to perform various preprocessing operations on these unstructured texts, including normalization, text cleaning to remove symbols and stop words, word segmentation correction, word embedding to generate word vectors, hot encoding, and labeling operations, so that these numerical or non-numerical text indicators can also be directly recognized and processed by the evaluator described in detail below.

[0050] That is, the signs of placenta accreta include: placental position, placental thickness, retroplacental hypoechoic zone, bladder line, placental crypt, placental base blood flow, cervical sinusoids and / or cervical morphology.

[0051] Furthermore, the multiple blood flow indices include: a vascular index, a blood flow index, a vascular blood flow index, a resistance index, a pulsatility index, and / or a ratio of end-systolic peak to end-diastolic peak, collected during seven time intervals of pregnancy: 11-16 weeks, 16-20 weeks, 20-24 weeks, 24-28 weeks, 28-32 weeks, 32-36 weeks, and 38 weeks and above. The resistance index, pulsatility index, and ratio of end-systolic peak to end-diastolic peak include the resistance index, pulsatility index, and ratio of end-systolic peak to end-diastolic peak of the umbilical artery, the right uterine artery, and the left uterine artery.

[0052] The placenta accreta sign evaluation device further comprises a screening unit 2, the process of which is to screen effective blood flow indicators as follows: Figure 2As shown, "Other" represents corresponding indicator data collected from patients with conditions other than placenta accreta, such as gestational diabetes mellitus (GDM), fetal growth restriction (TGR), and pregnancy-induced hypertension (PIH). Valid blood flow indicators for placenta accreta sign classification are extracted from these multiple blood flow indicators using the first mode and / or (preferably, or, as the number of valid blood flow indicators obtained by intersection is often too small when inputting a small sample size) the second mode, and these valid blood flow indicators are input into the fusion unit. Statistical methods based on intergroup difference analysis are well suited for distinguishing clinical medical diagnostic indicators. They strictly adhere to hypothesis testing, effectively avoid incorrect classifications, and are well-suited for small sample analysis. Machine learning methods, on the other hand, are more suitable for situations with a large number of features and nonlinear features. From the above, it can be seen that the blood flow indicators finally output by the acquisition unit of the present application to the screening unit can be 12n. For example, when the interval is n=7, the blood flow indicators will reach 84. In this case, using a machine learning classifier to extract effective blood flow indicators is a feasible solution; in addition, many classifiers (such as logistic regression and support vector machines) are interpretable and can also help medical personnel understand why the screened blood flow indicators can effectively classify signs of placenta accreta.

[0053] Therefore, the first mode and the second mode can respectively use machine learning classifiers and statistical analysis to process and extract effective indicators from the numerous blood flow indicators described in this application. Only the common blood flow indicators screened out in the two modes can be retained, reducing the situation where, for example, false positive indicators are screened out. The effective blood flow indicators obtained in this way have strong evidence support and can ensure the accuracy of placenta implantation classification within a limited range. Another screening method is to retain all blood flow indicators screened out in the two modes to ensure the integrity of blood flow information and avoid missing weakly correlated blood flow features. In addition, the effective blood flow features obtained in the two modes can be fused to balance the rigor and flexibility of the blood flow indicators. For example, the normalized feature importance value (such as the feature weight coefficient of SVM) and the significance value (such as the significance level p) of the effective blood flow indicators screened out in the first mode and the second mode are used as the weight value of the effective blood flow indicators to weightedly fuse these effective blood flow indicators.

[0054] Specifically, in one example of the present application, the first mode employs a machine learning algorithm, utilizing blood flow indicators collected starting at 11 weeks of gestation to train a model to fit the data and identify patterns in the indicators' classification of signs of placenta accreta. Multiple blood flow indicators are calculated and preprocessed in the acquisition unit before being input into the screening unit. Each of the multiple blood flow indicators has two labels: asymptomatic or with signs of placenta accreta. A support vector machine (SVM) classifier is configured in the screening unit, utilizing kernel techniques for mapping raw data into a high-dimensional space. This classifier is capable of handling nonlinear features and exhibits high training efficiency when the sample size is small. Because the nonlinear relationship between blood flow parameters and signs of placenta accreta is strong, three endometrial volumetric blood flow parameters (VI, FI, and VFI) from different gestational stages, three blood flow monitoring and analysis parameters (RI, PI, and S / D), or all six parameters can be selected as blood flow indicators to be screened and input into the SVM. The kernel function of SVM can select Gaussian kernel (RBF) to deal with the complex nonlinear relationship between blood flow indicators, especially the nonlinear relationship at the classification boundary, or select polynomial kernel to set the degree of nonlinearity to determine the optimal hyperplane shape for separating the blood flow indicators of two labels.

[0055] Then, the selected blood flow indicators are input into the SVM classifier, and the number of blood flow indicators is optimized so that the SVM achieves the preset classification performance, such as its accuracy (reflecting the overall classification performance of signs and asymptomatic), recall rate (reflecting the ability to identify signs), precision rate (avoiding misjudging asymptomatic as signs), etc. to reach the preset values. You can choose to split the multiple blood flow indicators in the training set evenly, use each split indicator to train the classifier separately and test its performance, record the performance under different combinations of indicator numbers and / or types (such as indicators of endometrial volume and indicators of blood flow monitoring and analysis), and obtain the combination of blood flow indicators that achieves the indicators or does not achieve the indicators but is the best through comparison, and set it as the effective blood flow indicator. The effective blood flow indicator obtained in this way can avoid overfitting caused by the classifier over-emphasizing part of the data, and can also ensure a certain classifier fitting efficiency when a small sample is input in a medical scenario.

[0056] In particular, a recursive approach can be used to gradually eliminate blood flow indicators with low importance in distinguishing signs of placenta accreta. The contribution can be represented by the feature weight coefficients in the SVM classifier. After each classifier training, the weight coefficients for each blood flow indicator are obtained, the blood flow indicator with the lowest weight coefficient is removed, the classifier is retrained, and the weight coefficients are repeatedly obtained. Based on the new classifier's placenta accreta classification performance, it is determined whether the blood flow indicator with the lowest weight coefficient should be further removed. For example, if the classifier's classification performance improved after removing the blood flow indicator with the lowest weight coefficient in the previous training round, then the blood flow indicator with the lowest weight coefficient could be considered for removal in the current training round. If the classifier's classification performance decreased after removing the blood flow indicator with the lowest weight coefficient in the previous training round, then the deep relationship between the removed indicator and other indicators should be considered. The removed indicator could be reintroduced and the blood flow indicators ranked after it could be removed, and the classifier could be retrained to verify the change in classification performance.

[0057] In this way, the classifier performance will not continue to improve after any remaining blood flow indicators are eliminated, so the remaining blood flow indicators can be determined to be valid blood flow indicators. In particular, in this first mode, it is preferred to set a minimum number of valid blood flow indicators to avoid excessive elimination of blood flow indicators, which may cause the model to overfit or suddenly underfit after a certain deletion. In one example, the first mode was used to construct a classifier (RBF-SVM) for three endometrial volume blood flow parameters. It was found that the classification weight values ​​of VI, FI and VFI were similar, there was no significant difference in the predictive performance of placenta accreta signs, and the prediction effect of the classifier on the test set reached 89%. However, in this example, the number of positive samples (i.e., samples with confirmed placenta accreta) in the training set and test set was too small, and there may be overfitting caused by data sparsity. Therefore, it is necessary to use the second mode to further screen valid blood flow indicators from these small sample data.

[0058] That is, the first mode includes: inputting the multiple blood flow indicators into the classifier, in response to the evaluation index value of the classifier not reaching the preset standard, adjusting the number and / or type of blood flow indicators input to the classifier, and in response to the evaluation index value of the classifier reaching the preset standard, setting the blood flow indicator currently input to the classifier as the valid blood flow indicator.

[0059] In addition, the first mode includes: inputting the multiple blood flow indicators into the classifier, and sorting the placenta accretion sign classification weight values ​​of the multiple blood flow indicators from low to high; eliminating blood flow indicators in order, retraining the classifier and sorting the multiple blood flow indicators from low to high based on the new weight values, in response to the increase of the classifier evaluation index value, continuing the classifier training, weight value sorting and sequential elimination of blood flow indicators until the classifier evaluation index value no longer increases, in response to the classifier evaluation index value decreasing, reintroducing the eliminated blood flow indicator and eliminating the next blood flow indicator in order, and continuing the classifier training, weight value sorting and evaluation index value comparison steps; and using the last retained blood flow indicator as the effective blood flow indicator.

[0060] In another example of the present application, the second mode uses a statistical intergroup analysis method to screen effective blood flow indicators. It can be considered to calculate the average and variance of blood flow indicators of similar samples to observe the relationship between indicators and signs from a holistic perspective. This mode is more stable and interpretable when the sample size of blood flow indicators is small. For example, intergroup analysis is performed on the three indicators of VI, FI, VFI, and RI, PI, and S / D of the umbilical artery, right uterine artery, and left uterine artery in different gestational time intervals.

[0061] Specifically, the mean and variance of the blood flow index data of the same type in each pregnancy were calculated, the mean was plotted as a bar graph, and the variance range was represented as a line segment for comparison. The results of the RI, PI and S / D indices of the umbilical artery are as follows: Figure 3 As shown, the mean values ​​of the umbilical artery RI, PI, and S / D indices all showed a decreasing trend with increasing gestation, and their variances also decreased. The umbilical artery RI data showed a small change, with some abrupt changes in variance, possibly due to adjustments in ultrasound equipment parameters during individual data collection. The mean value of the umbilical artery PI data decreased significantly in pregnancies with signs of placenta accreta, while the decrease was less pronounced in pregnancies without symptoms. The umbilical artery S / D index showed the most significant change, with the mean value in pregnancies with signs of placenta accreta decreasing the most, and the magnitude of the decrease was greater than that in pregnancies without symptoms.

[0062] The blood flow indicators of the right uterine artery include the right uterine artery RI value, the right uterine artery PI value and the right uterine artery S / D value. The sample categories are still the two types of placenta implantation signs and asymptomatic. The data of the right uterine artery blood flow indicators are shown in the figure below. Figure 4 As shown in Figure 2, the mean and variance of right uterine artery blood flow indices gradually decreased over the course of pregnancy, with the S / D index of the right uterine artery decreasing most significantly. Over time, the mean of each symptom-related data point for the right uterine artery remained lower than that of normal pregnant women and gradually decreased over time, but the difference in the decrease between the two groups remained consistent.

[0063] The left uterine artery observation indicators include the left uterine artery RI value, the left uterine artery PI value and the left uterine artery S / D value. The indicator classification is the same as above. The left uterine artery blood flow indicator data is shown in the figure below. Figure 5 The downward trend of the left uterine artery data is similar to that of the right uterine artery data. However, it is worth noting that the left uterine artery S / D value of the symptomatic samples is significantly lower than that of the asymptomatic samples in the early pregnancy (11-16 weeks), but is basically consistent with that of the asymptomatic samples in the late pregnancy (28 weeks and beyond), and the change range is lower than that of the asymptomatic samples.

[0064] Finally, for the three endometrial volume and blood flow parameters, the data graphs are as follows: Figure 6 As shown in the figure, it can be seen that the reduction values ​​of VI and VFI in the symptomatic samples throughout the pregnancy period are significantly lower than the reduction values ​​of asymptomatic samples; in addition, for each time interval, except for 11-16 weeks and 28-32 weeks, the numerical means of the two types of samples are relatively close, which makes the data difference small and the distinction more difficult.

[0065] From the above, we can see that for the changes in blood flow indicators throughout pregnancy as a classification of signs of placenta accreta and asymptomatic signs, the changes in umbilical artery RI between 16 and 38 weeks of pregnancy, the changes in left uterine artery S / D between 11 and 32 weeks of pregnancy, and the changes in endometrial volume VI and VFI between 11 and 38 weeks of pregnancy can be selected as effective blood flow indicators. However, in reality, not all subjects can provide blood flow indicator changes over such a long time span. In particular, for pregnant women with a short pregnancy, the aforementioned effective blood flow parameters are difficult or impossible to obtain effectively. Therefore, finding effective blood flow parameters within each gestational time interval is a more practical examination method that can cover more subjects, that is, using the values ​​of blood flow indicators in each gestational period as a classification of signs of placenta accreta and asymptomatic signs. In this case, the left uterine artery PI and left uterine artery S / D at 11-16 weeks of pregnancy; the umbilical artery S / D at 16-20 weeks of pregnancy; and the left uterine artery PI and left uterine artery S / D at 24-28 weeks of pregnancy can be selected as effective blood flow indicators. However, VI, FI and VFI are difficult to be used as effective blood flow indicators within a single interval.

[0066] That is, the second mode includes: performing inter-group analysis of the placenta accreta signs classification under multiple time intervals on the multiple blood flow indicators to determine that the blood flow indicator with significant difference between patients with placenta accreta signs and healthy people among the multiple blood flow indicators is the effective blood flow indicator.

[0067] The placenta accreta sign assessment device further includes a fusion unit 3 that inputs the effective blood flow index and the multiple clinical indicators into respective evaluators and fuses the placenta accreta sign classification results output by the evaluators to obtain a placenta accreta sign assessment result. The evaluators, each trained using the effective blood flow index and the multiple processed clinical indicators, can be fused using a weighted average approach. The weights of the evaluators can be set based on the performance of each classifier on its validation set. For example, each classifier is assigned a weight based on the normalized values ​​of its accuracy, recall, precision, or AUC, and then weighted fusion is performed using the following formula:

[0068]

[0069] Where P(y|x) is the predicted probability of placenta accreta signs and asymptomatic signs for a sample under multiple evaluators, and the category with the highest probability is selected as the placenta accreta sign classification result of the sample:

[0070]

[0071] When the SVM classifier is used as the evaluator, it can be subjected to equal regression to output the classification probability value, and then weighted fusion can be performed; when the logistic regression is used as the evaluator, weighted fusion can be performed directly based on the probability value it outputs.

[0072] In addition, the fusion of multimodal indicators can be achieved by training a new evaluator using evaluators of different modal indicators. Specifically, the outputs of the evaluators trained with the two modal indicators (same as above, preferably the probability value of the classification result) are spliced ​​into a new indicator to train a new evaluator:

[0073] φ(x)=[P1(y|x),P2(y|x)]

[0074] (Formula 3)

[0075]

[0076] In particular, the process of combining the outputs of the evaluators trained using the two modal indicators is preferably to split all effective blood flow indicators and clinical indicators into multiple groups, train their respective evaluators through cross-training, and then combine the probability values ​​output by the evaluators to form a new indicator matrix φ(x). This can prevent overfitting of the evaluators for the effective blood flow indicators and clinical indicators. It is understood that the evaluator for the effective blood flow characteristics can directly use the machine learning classifier used to screen them in the first mode. This machine learning classifier has been cross-validated in the first mode and can be directly used as the evaluator to generate the probability value of placenta accreta classification.

[0077] That is, fusing the placenta accreta sign classification results output by the evaluator to obtain a placenta accreta sign evaluation result includes: inputting the multiple effective blood flow indicators and the multiple clinical indicators into the first evaluator and the second evaluator, respectively, to obtain a first classification result and a second classification result, respectively; assigning a first weight and a second weight to the first classification result and the second classification result according to the evaluation index values ​​of the first evaluator and the second evaluator, respectively, to weight the first classification result and the second classification result to obtain the placenta accreta sign evaluation result.

[0078] Furthermore, fusing the placenta accreta sign classification results output by the evaluator to obtain a placenta accreta sign classification result includes: inputting the multiple effective blood flow indicators and the multiple clinical indicators into a first evaluator and a second evaluator, respectively, to obtain a first classification result and a second classification result, respectively; inputting the first classification result and the second classification result into a third evaluator, training the third evaluator so that its evaluation index values ​​reach preset standards, and outputting a third classification result from the third evaluator as the placenta accreta sign evaluation result.

[0079] In this way, the results of the placenta accreta sign classification are obtained through the above units, as well as the effective blood flow indicators required for the classification and a combination of clinical indicators fused with the effective blood flow indicators.

[0080] Based on these results, the present application further provides a multimodal placenta accreta assessment system, comprising a sign prediction unit and an evaluation unit. The sign prediction unit is configured to collect multiple clinical indicators and multiple blood flow indicators from a subject, fuse these multiple clinical indicators and multiple blood flow indicators, and predict the classification of each of the subject's multiple placenta accreta signs based on the fused indicators, and input the prediction results into the evaluation unit. It will be readily apparent that the sign prediction unit is the placenta accreta assessment device described above that determines the required indicators. It may not include the screening unit of the placenta accreta assessment device, but may also include a screening unit to regularly and dynamically update the effective blood flow indicators, achieving dynamic updating of the entire system. The system then automatically generates an accurate classification of placenta accreta signs by processing the input effective blood flow indicators and clinical indicators.

[0081] The assessment unit is configured to assign a third weight to each of the multiple signs of placenta accreta, receive the prediction results, and weightedly fuse the multiple signs of placenta accreta to obtain a placenta accreta assessment result for the subject. It is understood that different signs of placenta accreta may have different degrees of severity in impacting the life and health of the subject. The multiple signs of placenta accreta output by the sign prediction unit can be used as new input features to train a new classifier. The severity of the subject's placenta accreta is used as a classification label to determine weight coefficients for the different signs of placenta accreta. The weight coefficients are then used to weight the multiple signs of placenta accreta, and the weighted result is used as the final placenta accreta severity assessment result.

[0082] In one example, the current clinical assessment of the severity of placenta accreta is obtained by equally weighting all signs of placenta accreta. This means that all signs are assumed to have the same or nearly the same ability to reflect the severity of placenta accreta, and therefore all third weights can be set to 1. However, as existing clinical placenta accreta diagnostic criteria are optimized, the present application can use the assessment unit to timely change the third weights of different signs of placenta accreta to provide subjects with more accurate placenta accreta assessment results.

[0083] The basic principles of the present application have been described above in conjunction with specific embodiments. However, it should be noted that the advantages, strengths, and effects mentioned in this application are merely illustrative and not restrictive, and it should not be assumed that these advantages, strengths, and effects are required of each embodiment of this application. In addition, the specific details disclosed above are merely illustrative and facilitating understanding, and are not restrictive. The above details do not limit this application to necessarily being implemented using the above specific details.

[0084] The block diagrams of the devices, devices, equipment, and systems involved in this application are merely illustrative examples and are not intended to require or imply that they must be connected, arranged, or configured in the manner shown in the block diagrams. As will be appreciated by those skilled in the art, these devices, devices, equipment, and systems can be connected, arranged, or configured in any manner. Words such as "include," "comprise," "have," and the like are open-ended words, meaning "including but not limited to," and can be used interchangeably therewith. The words "or" and "and" used herein refer to the words "and / or" and can be used interchangeably therewith, unless the context clearly indicates otherwise. The word "such as" used herein refers to the phrase "such as but not limited to," and can be used interchangeably therewith.

[0085] It should also be noted that in the apparatus, device, and method of the present application, each component or each step can be decomposed and / or recombined, and such decomposition and / or recombination should be regarded as equivalent solutions of the present application.

[0086] The above description of the disclosed aspects is provided to enable any person skilled in the art to make or use the present application. Various modifications to these aspects will be readily apparent to those skilled in the art, and the general principles defined herein may be applied to other aspects without departing from the scope of the present application. Therefore, the present application is not intended to be limited to the aspects shown herein, but rather to be accorded the widest scope consistent with the principles and novel features disclosed herein.

[0087] The above description has been provided for the purpose of illustration and description. Furthermore, this description is not intended to limit the embodiments of the present application to the forms disclosed herein. Although a number of example aspects and embodiments have been discussed above, those skilled in the art will recognize certain variations, modifications, alterations, additions, and sub-combinations thereof.

Claims

1. A device for assessing signs of placenta accreta, comprising: an acquisition unit, which acquires and pre-processes a plurality of blood flow indices and a plurality of clinical indices, inputs the pre-processed plurality of blood flow indices into a screening unit, and inputs the plurality of clinical indices into a fusion unit; a screening unit, extracting effective blood flow indicators for placenta accreta sign classification from the plurality of blood flow indicators using the first mode and / or the second mode, and inputting the effective blood flow indicators into a fusion unit; A fusion unit inputs the effective blood flow index and the multiple clinical indexes into respective evaluators, and fuses the placenta accreta sign classification results output by the evaluators to obtain a placenta accreta sign evaluation result.

2. The placenta accreta sign assessment device according to claim 1, wherein: The first mode includes: The multiple blood flow indicators are input into the classifier. In response to the evaluation indicator value of the classifier not reaching the preset standard, the number and / or type of the blood flow indicators input into the classifier are adjusted. In response to the evaluation indicator value of the classifier reaching the preset standard, the blood flow indicator currently input into the classifier is set as the valid blood flow indicator.

3. The placenta accreta sign assessment device according to claim 1, wherein: The first mode includes: Inputting the multiple blood flow indicators into a classifier, and sorting the placenta accreta sign classification weight values ​​of the multiple blood flow indicators from low to high; Eliminating blood flow indicators in order, retraining the classifier and sorting the multiple blood flow indicators from low to high based on the new weight values, in response to an increase in the classifier evaluation index value, continuing the classifier training, weight value sorting and sequentially eliminating blood flow indicators until the classifier evaluation index value no longer increases, in response to a decrease in the classifier evaluation index value, reintroducing the eliminated blood flow indicator and sequentially eliminating the next blood flow indicator, and continuing the classifier training, weight value sorting and evaluation index value comparison steps; The blood flow index retained last is used as the effective blood flow index.

4. The placenta accreta sign assessment device according to claim 1, wherein: The second mode includes: An intergroup analysis of the placenta accreta signs classification under multiple time intervals is performed on the multiple blood flow indices to determine that the blood flow indices with significant differences between patients with placenta accreta signs and healthy people are the effective blood flow indices.

5. The placenta accreta sign evaluation device according to claim 1, wherein: Fusing the placenta accreta sign classification results output by the evaluator to obtain a placenta accreta sign assessment result includes: Inputting the multiple effective blood flow indicators and the multiple clinical indicators into a first evaluator and a second evaluator, respectively, to obtain a first classification result and a second classification result; A first weight and a second weight are assigned to the first classification result and the second classification result according to the evaluation index values ​​of the first evaluator and the second evaluator, respectively, so as to weight the first classification result and the second classification result to obtain the placenta accreta sign evaluation result.

6. The placenta accreta sign evaluation device according to claim 1, wherein: Fusing the placenta accreta sign classification results output by the evaluator to obtain the placenta accreta sign classification results includes: Inputting the multiple effective blood flow indicators and the multiple clinical indicators into a first evaluator and a second evaluator, respectively, to obtain a first classification result and a second classification result; The first classification result and the second classification result are input into a third evaluator, the third evaluator is trained so that its evaluation index value reaches a preset standard, and the third classification result output by the third evaluator is used as the placenta accreta sign evaluation result.

7. The placenta accreta sign assessment device according to any one of claims 1 to 6, wherein: The signs of placenta accreta include: placental position, placental thickness, retroplacental hypoechoic zone, bladder line, placental crypt, placental base blood flow, cervical sinusoids and / or cervical morphology.

8. The placenta accreta sign assessment device according to any one of claims 1 to 6, wherein: The multiple blood flow indicators are collected during seven time intervals of pregnancy: 11-16 weeks, 16-20 weeks, 20-24 weeks, 24-28 weeks, 28-32 weeks, 32-36 weeks, and 38 weeks and above: Vascular index, blood flow index, vascular blood flow index, resistance index, pulsatility index and / or ratio of peak end-systolic to peak end-diastolic pressure.

9. The placenta accreta sign assessment device according to claim 8, wherein: The resistance index, the pulsatility index and the ratio of end-systolic peak to end-diastolic peak include the resistance index, pulsatility index and ratio of end-systolic peak to end-diastolic peak of the umbilical artery, the right uterine artery and the left uterine artery respectively.

10. Multimodal placenta accreta assessment system, including: a sign prediction unit that collects multiple clinical indicators and multiple blood flow indicators of the subject and fuses the multiple clinical indicators and the multiple blood flow indicators, predicts the type of each placenta accreta sign among the multiple placenta accreta signs of the subject based on the fused indicators, and inputs the prediction result into the evaluation unit; An evaluation unit assigns a respective third weight to each of the multiple placenta accreta signs, receives the prediction result, and weightedly fuses the multiple placenta accreta signs to obtain a placenta accreta evaluation result of the subject.

Images