Multi-modal brain image and tryptophan metabolism characteristic-based panic disorder diagnosis system

Abstract

The invention discloses a panic disorder diagnosis system based on multi-mode brain images and tryptophan metabolism characteristics, and belongs to the technical field of medical information processing. A panic disorder diagnosis product comprises a tryptophan metabolism marker detection device, and tryptophan metabolism markers comprise one or more of 5-hydroxytryptophan, 5-hydroxytryptophan, kynurenine, 3-hydroxykynurenine, 3-hydroxyanthranilic acid, kynuric acid, quinolinic acid and picolinic acid. According to the method, key characteristics of panic disorder multi-mode brain images and tryptophan metabolism can be better captured, the efficiency of early diagnosis of diseases is remarkably improved, and an objective quantitative standard with high interpretability is provided for early screening of the diseases.

Description

Technical Field

[0001] The present invention relates to the field of medical information processing technology, and in particular to a panic disorder diagnosis system based on multimodal brain imaging and tryptophan metabolic characteristics. Background Art

[0002] Panic disorder (PD), a common clinical mental disorder, is a typical type of anxiety disorder with a high incidence, a prolonged course, and recurrent attacks. It causes great physical and mental pain to patients, leading to long-term social impairment and mental disability. The lifetime prevalence of PD is 1% to 5%, and the incidence in the general population is 3.4% to 4.7%. According to the International Classification of Diseases, 11th Revision (ICD-11) (WHO, 2022), PD is classified as a common chronic disease with a significant impact on patients' quality of life and socioeconomic status.

[0003] The most notable feature of PD is repeated panic attacks, and patients usually have symptoms such as increased heart rate, chest pain, and difficulty breathing. Studies have shown that cardiopulmonary symptoms are particularly prominent in PD patients. Among patients with cardiopulmonary symptoms as their main complaint, nearly half have panic-related problems. According to statistics, approximately 30%-50% of PD patients are misdiagnosed with cardiovascular disease during their first attack. PD patients repeatedly visit the emergency room, occupying a large amount of precious medical resources, delaying the disease, and having a poor prognosis. Therefore, early identification and diagnosis are crucial. However, the current diagnosis of PD mainly relies on symptomatic evaluation and lacks objective biological indicators, which makes the clinical diagnosis of PD very challenging.

[0004] Recent studies have shown that tryptophan metabolism may bidirectionally regulate the brain through the HPA axis, immune inflammation, and imbalances between neurotoxicity and neuroprotection. However, existing PD research is largely limited to single-modality brain imaging features, lacks comprehensive analysis of tryptophan pathway metabolites, and fails to integrate and analyze these multidimensional data. Therefore, systematically analyzing the characteristics of brain and tryptophan metabolism and constructing a multidimensional biomarker system are of great value in advancing PD diagnosis from symptomatology to biological assessment and diagnosis. Summary of the Invention

[0005] In response to the shortcomings of the existing technology, the present invention proposes a panic disorder diagnosis system based on multimodal brain imaging and tryptophan metabolic characteristics.

[0006] The purpose of the present invention can be achieved through the following technical solutions:

[0007] The first aspect of the present invention relates to a panic disorder diagnostic product, comprising: a tryptophan metabolic marker detection device, wherein the tryptophan metabolic marker includes one or more of 5-hydroxytryptamine, 5-hydroxytryptophan, kynurenine, 3-hydroxykynurenine, 3-hydroxyanthranilic acid, kynurenic acid, quinolinic acid, and picolinic acid.

[0008] Optionally, a multimodal brain image feature extraction device is also included, and the multimodal brain image features include: one or more of brain structure features, white matter fiber structure network features, gray matter static functional network features, white matter static functional network features and white matter dynamic functional network features.

[0009] Optionally, the brain structure features include gray matter thickness, white matter FA, white matter RD and white matter MD; the white matter fiber structure network features include: global properties of the white matter fiber structure network.

[0010] Optionally, the gray matter static functional network features include gray matter static topological features aNDc and gray matter static topological features aNLp.

[0011] The white matter static functional network features include white matter static topological features aNe, white matter static topological features aNDc and white matter static topological features aNBc.

[0012] The white matter dynamic functional network features include white matter dynamic topological features aNLe and white matter dynamic topological features aNCp.

[0013] Optionally, the gray matter static functional network features, white matter static functional network features, and white matter dynamic functional network features are collected from:

[0014] left thalamus;

[0015] right thalamus;

[0016] right superior longitudinal fasciculus;

[0017] right cingulate bundle;

[0018] right anterior corona radiata;

[0019] right posterior thalamic radiation;

[0020] left sagittal layer;

[0021] right anterior limb of the internal capsule;

[0022] right superior fronto-occipital fasciculus;

[0023] left sagittal layer;

[0024] right superior longitudinal fasciculus;

[0025] and, right superior fronto-occipital fasciculus.

[0026] Optionally, the white matter fiber structural characteristics are collected at the fornix fasciculus, and the gray matter thickness is collected at the right precentral gyrus.

[0027] A second aspect of the present invention relates to a method for constructing a panic disorder diagnostic model, comprising the following steps:

[0028] Acquiring a data set containing brain structure and functional networks and / or tryptophan metabolic features; the tryptophan metabolic features include one or more of tryptophan, 5-hydroxytryptamine, 5-hydroxytryptophan, kynurenine, 3-hydroxykynurenine, 3-hydroxyanthranilic acid, kynurenic acid, quinolinic acid, picolinic acid, the ratio of kynurenic acid to kynurenine, the ratio of kynurenic acid to tryptophan, the ratio of kynurenic acid to quinolinic acid, the ratio of kynurenic acid to 3-hydroxykynurenine, the ratio of kynurenic acid to 3-hydroxyanthranilic acid, and the ratio of picolinic acid to quinolinic acid; the brain structure and functional network features include one or more of brain structure features, white matter fiber structure network features, gray matter static functional network features, white matter static functional network features, and white matter dynamic functional network features;

[0029] A random forest classification prediction model was constructed using machine learning methods, and the data set was divided into a training set and a test set;

[0030] Perform model training and evaluation on the training set;

[0031] The model performance is evaluated on the test set to obtain the prediction model.

[0032] Optionally, 10-fold cross validation repeated 10 times is used to perform model training and evaluation on the training set.

[0033] A third aspect of the present invention relates to a panic disorder diagnostic model constructed by the above-mentioned method for constructing a panic disorder diagnostic model.

[0034] The fourth aspect of the present invention relates to reagents for detecting tryptophan metabolic markers, multimodal brain imaging features or the above-mentioned panic disorder diagnostic model, and their use in preparing panic disorder diagnostic products, wherein the tryptophan metabolic markers include one or more of 5-hydroxytryptophan, 5-hydroxytryptophan, kynurenine, 3-hydroxykynurenine, 3-hydroxyanthranilic acid, kynurenic acid, quinolinic acid, and picolinic acid; the multimodal brain imaging features include one or more of brain structural features, white matter fiber structural network features, gray matter static functional network features, white matter static functional network features, and white matter dynamic functional network features.

[0035] A panic disorder diagnosis system based on multimodal brain imaging and tryptophan metabolism characteristics, including: a data acquisition module, a preprocessing module, a brain network construction and graph theory-based network topology attribute analysis module, a feature extraction module, and a diagnosis module;

[0036] The data collection module collects demographic information, brain imaging, and tryptophan-related metabolite concentration level data of healthy controls or patients with panic disorder;

[0037] The preprocessing module preprocesses the brain imaging data collected by the data acquisition module, including: T1-weighted structural image preprocessing, DTI image preprocessing, and BOLD-fMRI image preprocessing;

[0038] The brain network construction and graph theory-based network topology property analysis module, wherein the brain network construction includes structural covariant network construction, white matter fiber-based structural network construction, gray matter static brain functional network construction, white matter static brain functional network construction, gray matter dynamic brain functional network construction, and white matter dynamic brain functional network construction; the graph theory-based network topology property analysis is to analyze the dynamic and static topological properties of gray matter and white matter functional networks. Under the premise of ensuring that the number of functional network connection edges is constant and the network has small-world properties, the sparsity threshold is selected from 0.01 to 0.34 with a step size of 0.01. The network topology properties include global indicators and local indicators;

[0039] The feature extraction module uses a data acquisition module, a preprocessing module, a brain network construction module, and a network topology attribute analysis module based on graph theory to obtain statistical analysis of brain structure indicators, brain network indicators, and tryptophan metabolism indicators with differences. The tryptophan-related indicators with differences are selected using support vector machine-recursive feature elimination (SVM-RFE) and a method of removing variables with strong correlations before entering the next step of analysis. The least absolute shrinkage and selection operator (LASSO) method is used to perform feature screening on the above-selected difference indicators, and the classification accuracy is used as the cost function of the model. Through LASSO regression, key features with non-zero coefficients are screened out.

[0040] The diagnostic module uses the sample data obtained by the feature extraction module to build a random forest (RF) classification prediction model. Random forest can integrate the results of multiple decision trees to improve model performance. 10-fold cross-validation (10-CV) is used on the training set for model training and evaluation. Finally, the selected features are used to evaluate the model performance on the test set.

[0041] Optionally, in the data acquisition module, demographic information includes gender, age, years of education, height, and weight; brain imaging data includes T1-weighted structural images, DTI images, and BOLD-fMRI images; the tryptophan-related metabolite concentration level data includes tryptophan (TRP), 5-hydroxytryptamine (5-HT), 5-hydroxytryptophan (5-HTP), kynurenine (KYN), 3-hydroxykynurenine (3-HK), 3-hydroxyanthranilic acid (3-HAA), kynurenic acid (KYNA), quinolinic acid (QUIN), picolinic acid (PICA), and KYNA / KYN, KYNA / TRP, KYNA / QUIN, KYNA / 3-HK, KYNA / 3-HAA, and PICA / QUIN ratios;

[0042] Optionally, the statistical analysis of the brain structure and brain network indicators and tryptophan-related metabolite indicators with differences is performed on the brain and tryptophan data of the panic disorder group and the healthy control group using age, gender, BMI and years of education as covariates to obtain difference indicators;

[0043] Optionally, the 10-fold cross-validation process is as follows: First, randomly divide the dataset into 10 subsets of similar size and balanced category ratio. Next, use one of the subsets as the test set and the remaining 9 subsets as the training set for model training. After model training is completed, use the trained model to predict the test set and calculate the corresponding performance indicators. This process is repeated 10 times, each time selecting a different subset as the test set. Finally, the results of the 10 validations are averaged to evaluate the performance of the model.

[0044] Optionally, the diagnostic module adopts a step-by-step verification framework, specifically including a two-stage diagnostic efficacy evaluation process:

[0045] Phase 1 (Full Image Feature Analysis): Based on the first dataset, a brain structural feature set, a brain functional network feature set, and a joint imaging feature set were constructed. The area under the receiver operating characteristic curve (ROC-AUC) of each feature set was calculated using a random forest model to verify the synergistic effect of structure-function combined diagnosis.

[0046] Phase II (combined imaging-metabolism analysis): Subsamples that completed tryptophan metabolism testing were selected from the Phase I dataset and the following operations were performed:

[0047] Step 1, feature reuse: extract the differential brain imaging indicators determined in the first stage;

[0048] Step 2, metabolic feature integration: differential tryptophan metabolites screened based on SVM-RFE;

[0049] Step 3, hierarchical performance verification: diagnostic performance of brain imaging features alone (reusing the first-stage model); diagnostic performance of tryptophan metabolites alone; diagnostic performance of the combined imaging-tryptophan metabolism feature set.

[0050] Beneficial effects of the present invention:

[0051] This study provides a method for diagnosing panic disorder based on physiological indicators. By extracting multimodal brain imaging features from subjects and integrating them with tryptophan metabolism data, the combined diagnostic value of metabolites and imaging markers is clarified. This method not only improves the objectivity of PD diagnosis but also provides a scalable technical approach for multi-omics biomarker research in psychiatric disorders. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] The present invention will be further described below with reference to the accompanying drawings.

[0053] Figure 1 It is the overall framework diagram of the present invention;

[0054] Figure 2 To construct a PD diagnosis and classification model based on gray / white matter structure and brain network topology features;

[0055] Figure 3 To construct a PD diagnosis and classification model by combining multimodal brain imaging features and tryptophan metabolism. DETAILED DESCRIPTION

[0056] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only 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 ordinary technicians in this field without making any creative efforts shall fall within the scope of protection of the present invention.

[0057] The terms involved in the embodiments of the present invention can be interpreted as follows:

[0058] Gray matter mainly includes neuronal cell bodies, dendrites, unmyelinated axon initial segments and glial cells (astrocytes, oligodendrocytes, etc.), and is rich in capillary networks.

[0059] White matter mainly includes myelinated axons (nerve fiber bundles), with a small amount of unmyelinated axons and oligodendrocytes.

[0060] LASSO (Least Absolute Shrinkage and Selection Operator) is a regularization and variable selection method for linear regression models. Based on the least squares estimation (OLS), L1 norm penalty term is introduced to compress some regression coefficients to 0, thereby achieving variable selection and preventing overfitting.

[0061] DTI is a magnetic resonance imaging (MRI) technique based on the diffusion motion of water molecules. It reveals the microstructure of white matter fiber bundles by measuring the directional dependence of water diffusion in brain tissue.

[0062] BOLD-fMRI (Blood Oxygenation Level Dependent fMRI) indirectly reflects brain functional activity by detecting local hemodynamic changes caused by neuronal activity.

[0063] In some embodiments of the present invention, a panic disorder diagnosis system and method based on multimodal brain imaging and tryptophan metabolism characteristics are disclosed, which may include the following steps:

[0064] Phase I: A total of 107 PD patients and 146 healthy controls (HCs) were enrolled;

[0065] 1. Collect brain imaging data

[0066] Demographic data and brain imaging data of HCs or PD patients were collected; demographic data included gender, age, years of education, height, and weight; brain imaging data included T1-weighted structural images, DTI images, and BOLD-fMRI images.

[0067] 2. Preprocessing

[0068] Including T1-weighted structural image preprocessing, DTI image preprocessing, and BOLD-fMRI image preprocessing:

[0069] T1-weighted structural images were preprocessed, including volume segmentation and cortical surface reconstruction. Based on the BNA-246 brain atlas (https: / / atlas.brainnetome.org / bnatlas.html), cortical thickness data for 105 cortical regions and volume data for 18 subcortical nuclei in each hemisphere were extracted.

[0070] DTI image preprocessing included extracting a brain mask and removing non-brain tissue, performing motion and eddy current correction, registration, and constructing a mean fractional anisotropy (FA) image and a mean FA skeleton. Combined with the standard white matter atlas developed by Johns Hopkins University (JHU) (JHU-ICBM-DTI-81), we extracted mean FA, mean diffusivity (MD), and radial diffusivity (RD) values for 48 white matter regions.

[0071] BOLD-fMRI image preprocessing includes data format conversion, removal of the first 10 time points, inter-slice time correction and head motion correction, registration, normalization, smoothing, linear drift removal, noise signal removal, and temporal bandpass filtering.

[0072] 3. Brain network construction and network topology attribute analysis based on graph theory

[0073] Brain network construction includes structural covariant network construction, white matter fiber-based structural network construction, gray matter static brain functional network construction, white matter static brain functional network construction, gray matter dynamic brain functional network construction, and white matter dynamic brain functional network construction. Network topology property analysis based on graph theory analyzes the dynamic and static topological properties of gray matter and white matter functional networks. Under the premise of ensuring that the number of functional network connection edges is constant and the network has small-world properties, the sparsity threshold is selected from 0.01 to 0.34 with a step size of 0.01. The network topology properties include global indicators and local indicators.

[0074] The structural covariation network was constructed based on the cortical thickness data of 210 brain regions extracted from the BNA-246 brain atlas. Linear regression analysis was performed on the morphological characteristics of each brain region, and age, gender, and BMI confounding factors were used as covariates. The residuals after the regression analysis were used for subsequent covariation relationship calculations. The Pearson correlation coefficient between the residuals of each brain region was calculated to obtain the structural covariation matrix (210×210) for each group of subjects. The elements in the covariation matrix with connection strength higher than the threshold were set to 1 (indicating the existence of connection), and the elements with connection strength lower than the threshold were set to 0 (indicating the absence of connection). Finally, a 210×210 binary matrix was obtained for subsequent graph theory analysis;

[0075] A structural network based on white matter fibers was constructed using the deterministic fiber assignment by continuous tracking (FACT) algorithm. Tracking was terminated if the MD was less than 0.2 or the turning angle exceeded 45°. The cortex was divided into 210 regions of interest (ROIs) based on the BNA-246 template, and each ROI was defined as a node in the network. A structural network matrix (210 × 210) was constructed for each subject by extracting the fiber numbers (FN) and MD values between two brain regions. To reduce false connections due to noise, the FN was set as a threshold of 3, and only edges with FN ≥ 3 were retained.

[0076] The gray matter static brain functional network was constructed based on the BNA-246 brain atlas. This study defined the nodes and edges of the brain network. First, the average time series of each brain region was extracted. The Pearson correlation coefficients between the time series of all brain regions were then calculated and converted to z-scores using the Fisher z-transform. Finally, a gray matter functional connectivity matrix (246 × 246) was constructed for each subject.

[0077] The white matter static brain functional network was constructed by defining brain network nodes and edges according to the JHU white matter template. The average time series of each white matter brain region was extracted, and the Pearson correlation coefficient between the time series of all brain regions was calculated and then converted into Fisher z-value to obtain the white matter functional connectivity correlation matrix (48 × 48) for each subject.

[0078] Gray matter dynamic brain functional networks were constructed using nodes defined in the BNA-246 brain atlas as ROIs. Dynamic functional connectivity between each pair of ROIs was calculated using a sliding time window method. Referring to previous studies, the sliding window parameters were set as follows: window width = 50 TR, step length = 2 TR. This generated a 91 × 246 × 246 dynamic functional connectivity matrix for each subject.

[0079] The white matter dynamic functional network was constructed using nodes defined by the JHU white matter template as ROIs. A sliding time window method was used to calculate the dynamic functional connectivity between each pair of ROIs. The sliding window parameters were the same as those used in the gray matter dynamic network construction, ultimately generating a 91×48×48 white matter dynamic functional connectivity matrix for each subject.

[0080] Global indicators of network topology properties include global efficiency (Eg), local efficiency (Eloc), and small-worldness. Small-worldness indicators include clustering coefficient (Cp), characteristic path length (Lp), normalized clustering coefficient (gamma), normalized characteristic path length (lambda), and small-worldness index (Sigma). Local indicators include nodal efficiency (Ne), nodal local efficiency (NLe), nodal clustering coefficient (NCp), nodal shortest path length (NLp), nodal degree centrality (NDc), and nodal betweenness centrality (NBc).

[0081] 4. Feature extraction

[0082] The brain structure and functional network data of the PD group and the HCs group were statistically tested with age, gender, BMI, and years of education as covariates. One-way covariance analysis was used to compare the differences between the groups (after FDR correction). The LASSO method was used to select features for the above-mentioned differential brain structure / functional network difference indicators, and the classification accuracy was used as the cost function of the model. Through LASSO regression, key features with non-zero coefficients were screened, as shown in Table 1 and Figure 2 The structural indicators obtained 1 gray matter thickness feature, 1 white matter FA feature, 1 white matter MD feature, and 3 white matter fiber structure network features. The functional network topology indicators obtained 2 gray matter static network features, 5 white matter static network features, and 5 white matter dynamic network features.

[0083] Table 1. Gray / white matter regions used to distinguish PD and HCs

[0084]

[0085]

[0086] 5. Diagnosis of panic disorder

[0087] A random forest classification prediction model was constructed using machine learning methods. The dataset was divided into a training set (80%) and a test set (20%). The selected brain structural and functional networks were used to evaluate model performance on the test set. To establish a reliable and stable classification model, 10-fold cross-validation was repeated 10 times on the training set for model training and evaluation. The steps were as follows: First, the dataset was randomly divided into 10 subsets of similar size and balanced class ratios. Second, one of the subsets was used as the test set, and the remaining nine subsets were used as the training set for model training. After model training, the trained model was used to predict the test set and calculate the corresponding performance indicators. This process was repeated 10 times, each time a different subset was selected as the test set. Finally, the results of the 10 validations were averaged to evaluate model performance. The selected features were then used to evaluate model performance on the test set. In order to evaluate whether the combination of multidimensional data sets is better than the lower dimensional data sets, in addition to using the combined feature set of brain structure (gray / white matter structure and structural network) and brain functional network topological attribute features (gray matter brain function dynamic / static network, white matter brain function dynamic / static network), separate feature sets of brain structure and brain function were constructed for the above performance evaluation. The evaluation index of model performance is ROC-AUC, such as Figure 2 The above 18 features were used to establish a random forest classification model. The results showed that the ROC curve analysis of brain indicators for distinguishing PD from HCs showed that the AUCs of brain structure and brain functional topological networks for predicting PD diagnosis were 0.725 and 0.777, respectively, while the AUC of the combined prediction of PD diagnosis was 0.851 (95% confidence interval = 0.7371-0.9630), suggesting that the random forest classification model constructed by combining these 18 gray / white matter structure and functional network features has a further improved AUC and can effectively distinguish PD from HCs.

[0088] Phase II: A sub-dataset based on the completed tryptophan metabolism test in the large sample of Phase I, including 78 PD patients and 94 HCs;

[0089] 1. Collect tryptophan metabolism data

[0090] The concentration level data of tryptophan-related metabolites include tryptophan (TRP), 5-hydroxytryptamine (5-HT), 5-hydroxytryptophan (5-HTP), kynurenine (KYN), 3-hydroxykynurenine (3-HK), 3-hydroxyanthranilic acid (3-HAA), kynurenic acid (KYNA), quinolinic acid (QUIN), picolinic acid (PICA) and the ratios of KYNA / KYN, KYNA / TRP, KYNA / QUIN, KYNA / 3-HK, KYNA / 3-HAA, and PICA / QUIN.

[0091] 2. Differential tryptophan metabolites screened based on SVM-RFE;

[0092] Comparisons of tryptophan metabolism between the PD and HC groups were statistically tested using age, sex, BMI, and years of education as covariates. One-way analysis of covariance was used to compare differences between the groups. The support vector machine-recurrent encephalogram (SVM-RFE) method was used to screen for these differentially expressed biomarkers. To reduce the impact of multicollinearity, variables with strong correlations were removed, ultimately retaining the features with the greatest potential for biomarker prediction. Specifically, if the correlation between two variables was high (using an absolute correlation coefficient ≥ 0.70 as the threshold), the variable with the higher average correlation was removed. Finally, four tryptophan metabolite ratios with biomarker potential were identified: PICA / QUIN, 3-HAA, QUIN, and KYNA / TRP.

[0093] 3. In this dataset, extract the differential brain imaging indicators determined in the first stage

[0094] The brain indicators that were found to have significant differences in the large sample analysis in the first stage were screened and extracted from the small sample data in this stage.

[0095] 4. Feature extraction

[0096] The LASSO method was used to select features of brain structure / function difference indicators (the selected brain structure and function indicators were all derived from brain indicators found to have significant differences through large-sample data analysis in the first phase to ensure the logical coherence and data consistency of the previous and subsequent phases of research) and tryptophan metabolites with biomarker potential, and the classification accuracy was used as the cost function of the model; through LASSO regression, key features with non-zero coefficients were screened out, as shown in Table 2 and Figure 3 , we obtained 1 gray matter thickness feature, 2 white matter microstructure features, 2 white matter fiber structure network features, 1 gray matter static network feature, 5 white matter static network features, 4 white matter static network features and 4 tryptophan indices.

[0097] Table 2. Multimodal brain imaging features and tryptophan-related metabolites for distinguishing PD from HCs

[0098]

[0099]

[0100] 5. Diagnosis of panic disorder

[0101] A machine learning method was used to construct a random forest classification prediction model, and the data set was divided into a training set (80%) and a test set (20%); the selected brain structure and functional network and the key features of tryptophan metabolism were used to evaluate the model performance in the test set; in order to establish a reliable and stable classification model, 10-fold cross-validation was repeated 10 times on the training set for model training and evaluation, and the steps were the same as the first stage. The selected features were then used to evaluate the model performance in the test set. In order to evaluate whether the combination set of multidimensional data is better than the lower dimensional data set, in addition to using the combined feature set of brain indicators and tryptophan metabolites, separate feature sets of brain indicators and tryptophan metabolites were also used to perform the above performance evaluation. The evaluation index of model performance is ROC-AUC, such as Figure 3 ,The 19 features mentioned above were used to establish a random forest classification model. ,The results showed that the AUCs for predicting PD diagnosis by brain ,image features and tryptophan metabolic indices were 0.874 and 0.921, ,respectively. The combined features could effectively distinguish PD and HCs with an AUC of 0.958 and a 95% ,credible interval of 0.8901-1.

[0102] In the technical solution of the present invention, the combination of two types of markers, brain imaging and tryptophan metabolic characteristics, can achieve a preliminary diagnostic effect. However, those skilled in the art should be aware that in the actual diagnosis process, it is not limited to using these two types of characteristics as the only diagnostic indicators. Other well-known diagnostic indicators and well-known symptoms can also be combined to obtain a more accurate diagnostic effect. In other words, the above-mentioned embodiments of the present invention do not imply that these two types of characteristics can only be used as separate indicators for diagnosis. In addition, the above-mentioned well-known diagnostic indicators and symptoms can be queried according to well-known technical information, and this application will not go into details.

[0103] Throughout this specification, references to terms such as "one embodiment," "example," or "specific example" indicate that the specific features, structures, materials, or characteristics described in conjunction with that embodiment or example are included in at least one embodiment or example of the present invention. In this specification, schematic representations of these terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples.

[0104] The basic principles, main features, and advantages of the present invention are shown and described above. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The above embodiments and descriptions are merely illustrative of the principles of the present invention. Various changes and modifications may be made to the present invention without departing from the spirit and scope of the present invention, and such changes and modifications fall within the scope of the invention as claimed.

Claims

1. A panic disorder diagnosis product, characterized in that: include: A tryptophan metabolic marker detection device, wherein the tryptophan metabolic marker includes one or more of 5-hydroxytryptamine, 5-hydroxytryptophan, kynurenine, 3-hydroxykynurenine, 3-hydroxyanthranilic acid, kynurenic acid, quinolinic acid, and picolinic acid.

2. The panic disorder diagnostic product according to claim 1, characterized in that: It also includes a multimodal brain image feature extraction device, and the multimodal brain image features include: one or more of brain structure features, white matter fiber structure network features, gray matter static functional network features, white matter static functional network features and white matter dynamic functional network features.

3. The panic disorder diagnostic product according to claim 2, characterized in that: The brain structure features include gray matter thickness, white matter FA, white matter RD and white matter MD; the white matter fiber structure network features include: the global properties of the white matter fiber structure network.

4. The panic disorder diagnostic product according to claim 2, characterized in that: The gray matter static functional network features include gray matter static topological features aNDc and gray matter static topological features aNLp; The white matter static functional network features include white matter static topological features aNe, white matter static topological features aNDc and white matter static topological features aNBc; The white matter dynamic functional network features include white matter dynamic topological features aNLe and white matter dynamic topological features aNCp.

5. The panic disorder diagnostic product according to claim 2, characterized in that: The collection sites of the gray matter static functional network features, white matter static functional network features and white matter dynamic functional network features include: left thalamus; right thalamus; right superior longitudinal fasciculus; right cingulate bundle; right anterior corona radiata; right posterior thalamic radiation; left sagittal layer; right anterior limb of the internal capsule; right superior fronto-occipital fasciculus; left sagittal layer; right superior longitudinal fasciculus; and, right superior fronto-occipital fasciculus.

6. The panic disorder diagnostic product according to claim 2, characterized in that: The site for collecting the white matter fiber structural characteristics is the fornix fasciculus; The gray matter thickness was collected from the right precentral gyrus.

7. A method for constructing a panic disorder diagnostic model, characterized in that: The following steps are involved: Acquiring a data set containing brain structural features and functional network features and / or tryptophan metabolic features; the tryptophan metabolic features include one or more of tryptophan, 5-hydroxytryptamine, 5-hydroxytryptophan, kynurenine, 3-hydroxykynurenine, 3-hydroxyanthranilic acid, kynurenic acid, quinolinic acid, picolinic acid, the ratio of kynurenic acid to kynurenine, the ratio of kynurenic acid to tryptophan, the ratio of kynurenic acid to quinolinic acid, the ratio of kynurenic acid to 3-hydroxykynurenine, the ratio of kynurenic acid to 3-hydroxyanthranilic acid, and the ratio of picolinic acid to quinolinic acid; A random forest classification prediction model was constructed using machine learning methods, and the data set was divided into a training set and a test set; Perform model training and evaluation on the training set; The model performance is evaluated on the test set to obtain the prediction model.

8. The method for constructing a panic disorder diagnostic model according to claim 1, wherein: Model training and evaluation were performed on the training set using 10-fold cross validation repeated 10 times.

9. A panic disorder diagnostic model constructed by the method for constructing a panic disorder diagnostic model according to claim 7 or 8.

10. Use of a reagent for detecting tryptophan metabolic markers, a multimodal brain imaging feature, or the panic disorder diagnostic model according to claim 9 in the preparation of a panic disorder diagnostic product, wherein the tryptophan metabolic markers include one or more of 5-hydroxytryptophan, 5-hydroxytryptophan, kynurenine, 3-hydroxykynurenine, 3-hydroxyanthranilic acid, kynurenic acid, quinolinic acid, and picolinic acid.

Images