CN114246588A

CN114246588A - Depression research method

Info

Publication number: CN114246588A
Application number: CN202111581763.6A
Authority: CN
Inventors: 何宗岭; 卢凤梅; 于跃; 陈勇; 岳玉川; 陈华富
Original assignee: Chengdu Fourth Peoples Hospital
Current assignee: Chengdu Fourth Peoples Hospital
Priority date: 2021-12-22
Filing date: 2021-12-22
Publication date: 2022-03-29

Abstract

The invention discloses a depression research method, which is characterized in that in the research process, the analysis is carried out on patients with unidirectional depression and bidirectional affective disorder in the depression attack stage and the analysis is used as a reference; in the clinical symptom evaluation process, factors combining the quality characteristic evaluation and the state characteristic evaluation are used as research indexes, and the method comprises the following implementation steps: image and clinical data acquisition → multi-modal brain network analysis → construction of a pattern classification prediction model. The invention has strong practicability and functionality, and can be widely applied to the technical field of depression research.

Description

Depression research method

Technical Field

The invention relates to the field of depression research, in particular to a depression research method.

Background

In the treatment process of the depression, the problems that the targeted treatment at a certain depth cannot be realized theoretically and microscopically, the depth analysis of the depression patient to a certain degree cannot be realized, and the improvement of a comprehensive and effective later-period treatment guidance scheme for the depression patient to a certain degree cannot be realized are solved.

Disclosure of Invention

In view of the above problems, the present invention proposes a novel method that can be studied for depression.

The technical scheme provided by the invention is as follows:

a method for studying depression, wherein during the course of the study, patients with unidirectional depression and bidirectional affective disorder in depressive episode are analyzed and used as reference; in the clinical symptom evaluation process, factors combining the quality characteristic evaluation and the state characteristic evaluation are used as research indexes, and the method comprises the following implementation steps:

image and clinical data acquisition → multi-modal brain network analysis → construction of a pattern classification prediction model.

Further, the image data acquisition comprises high spatial resolution structure data acquisition, resting state functional resonance data acquisition and diffusion tensor data acquisition;

the image data analysis comprises structural morphology analysis, functional network construction, structural network construction, pairwise spatial clustering, a resting state dynamic functional network and a depression pattern recognition classification model.

Further, in the structural morphology analysis process, a symmetrical average whole brain template is created, an grey matter template is created, a mirror image is created for each tested image, the tested original image and the created mirror image are registered to the generated template to be segmented, two voxel-based sample tests are carried out on the grey matter density/volume of two groups of data of the original image and the mirror image to obtain a parameter statistical graph, and a region with P <0.05 as a difference is obtained after multiple comparison and correction.

Further, in the functional network construction process, an area of interest is selected based on prior assumption or structural image analysis results;

extracting the average time sequence of all voxels in each interest region as a time sequence signal of the interest region;

performing time domain preprocessing on the time sequence signal of each interested region, wherein the time domain preprocessing comprises the step of returning a tested head motion signal; performing time domain preprocessing on the time sequence signals subjected to the band-pass filtering and removing the machine noise and the physiological noise, wherein the time domain preprocessing comprises the step of returning a tested head motion signal; removing machine noise and physiological noise by band-pass filtering, then calculating correlation coefficients for the interested region time sequence between every two adjacent interested regions, and normalizing all the correlation coefficients to normal distribution by using Fisher r-z transformation for the obtained correlation coefficients; each interested area is used as a node in the network, and the correlation coefficient between the two nodes is used as an edge between the two nodes to obtain the whole function network related to the depression;

for the functional network, obtaining the total connectivity of each node by adopting the connectivity;

statistically comparing the change of the node degree of the brain network of the patient and the normal tested brain network, searching physiological markers of the patient, and exploring the correlation between abnormal brain structure connection and quality characteristics and state characteristics of the depression.

Further, in the process of constructing the functional network, a whole brain functional connection network is adopted for construction, and in the process of constructing the structural network, diffusion tensor imaging is specifically included.

Further, in the pattern recognition classification model of depression, local indexes and connection indexes of the tested functions and the structural brain regions are extracted, and for each feature, the correlation with the depression scale is calculated.

Compared with the prior art, the invention has the advantages that:

by adopting the technical scheme provided by the invention, a multi-mode image analysis method can be developed, and the function of extracting the characteristic information of brain functions and structural activities is realized;

by adopting the method provided by the invention, the brain function structure and the brain network neural mechanism of the emotional neural circuit of the depression are detected;

by adopting the method provided by the invention, a key method for classifying and predicting development modes is adopted to realize early diagnosis and identification of depression.

Detailed Description

The present invention will be described in further detail with reference to examples.

(I) the subject

(1) Patient groups: the study project is intended to include 150 patients with depression at the onset of depression (unipolar depression) and bipolar disorder (bipolar depression), both from outpatients or inpatients at the clinical hospital of the institute of brain science, university of electronic technology (fourth national hospital in metropolis). All patients received two Clinical interviews with experienced psychiatrists and were screened using a Patient's version of the Clinical scheduled meeting sheet (SCID-P). The severity of the clinical symptoms in patients was assessed using the 24 Hamilton Depression Scale (HAMD-24) and the 14 Hamilton anxiety Scale (HAMA-14). Other clinical profile data related to the patient's depressive episodes were also collected, including age of first onset, overall course, number of depressive episodes, duration of single depressive episode, and drug load index (media load index).

Depression group:

grouping standard: age 18 to 60 years; (ii) compliance with the Diagnostic criteria for major depressive disorder in the United states of America Diagnostic and Statistical Manual, Fourth Edition, DSM-IV; ③ is currently in the depressive episode; and the cultural education degree is more than 9 years.

Exclusion criteria: combining schizophrenia, bipolar disorder, anxiety disorder or other serious mental disorder, having mental development retardation, having any history of consciousness loss and serious somatic or nervous system diseases, and having any drug dependence or abuse; ② pregnant women and nursing women; and thirdly, MRI contraindications exist, or the brain structure is found to be abnormal by MRI examination.

Group of bipolar disorders:

grouping standard: age 18 to 60 years; (ii) compliance with the Diagnostic criteria for bipolar affective disorder of the American Manual of mental and Statistical management, Fourth Edition, DSM-IV; ③ is currently in the depressive episode; and the cultural education degree is more than 9 years.

Exclusion criteria: combining schizophrenia, anxiety disorders or other severe mental disorders, having mental retardation, having any history of loss of consciousness, severe physical or neurological diseases, having any drug dependence or abuse; ② pregnant women and nursing women; and thirdly, MRI contraindications exist, or the brain structure is found to be abnormal by MRI examination.

(2) Control group: 200 healthy controls were recruited from the community by way of poster advertising and were screened using a non-patient version of the structured clinical stress session table (SCID-NP). Healthy controls were determined to have no history of severe somatic or neuropsychiatric disease, nor in their immediate relatives had psychiatric or neurological disease.

The study was approved by the ethics committee of the fourth people's hospital in metropolis, and all subjects were fully informed of the study and signed written informed consent. All subjects were tested by two experienced radiologists and no significant abnormalities were found in the conventional MRI tests before entering further studies.

(II) evaluation of clinical symptoms

(1) Estimation of predisposition characteristics

The Affective Neuroscience Personality Scale (ANPS) was used to assess predisposition characteristics associated with depression, which was a total of 112 subjects, primarily for assessing the emotional Personality architecture of individuals. Because of cultural and religious differences between the traditional and the western, social emotional tendencies, only 6 basic emotional personality factors of the original scale were used in the study. The 6 core emotional personality factors are exploration, fear, love, anger, entertainment and sadness respectively. The six basic emotion personality factors belong to two higher-level emotion personality dimensions, namely a positive emotion personality dimension and a negative emotion personality dimension. The positive emotional personality dimension includes exploration, care and entertainment, and the negative emotional personality dimension includes fear, anger, sadness. The scale adopts a Likert 4 point scale evaluation method, which is 1-4. And 14 scoring items for each core emotion factor, wherein seven items are suitable for the reverse scoring rule. For example, the exploration factor, wherein the seven items are based on the positive vocabulary query entry scoring criteria: 4, 3, 2, 1; correspondingly, the other seven items of item scoring criteria in the negative vocabulary query are as follows: 1, 2, 3, and 4.

(2) State feature evaluation

Loss of pleasure (Anhedonia) is one of the most representative core features of depression, considered as a potential phenotypic marker. Whether in clinical manifestation or laboratory measurement, the general expression of depression patients is higher than the loss of pleasure of normal people. Such as: patients with depression often self-report little or no pleasurable experiences, behaving with suppressed facial expressions, sounds, and limb movements. The study evaluated the pleasurable experience and the ability to experience Pleasure in subjects using the SHAPS Pleasure loss Scale (SHAPS). The SHAPS is a questionnaire containing 14 items of content, intended to reflect the degree of lack of pleasure by investigating the ability of the subject to enjoy pleasure in various situations.

Reduced Positive and increased Negative emotions are considered to be the pathological emotional feature of the core of depression, and the Positive and Negative affective Scale (PANAS) was used in this study to assess all core affective symptoms tested. The scale is a self-rating scale containing 20 items, and the Likert 5 point scale rating method is adopted, wherein the scales respectively represent from 1 to 5 that the scale is from 'none' to 'very strong'. The scale scores are combined into two sentiment dimensions, evaluating positive sentiment and negative sentiment respectively. Negative emotions (NA) mainly represent emotional features such as pain, anxiety, depression, etc., while positive emotions (PA) mainly include features such as pleasure, energetic and alert feelings.

(III) image data acquisition and processing

(1) Magnetic resonance data acquisition

MRI data were collected at the university of electronic technology medical information center, 3T GE Discovery MR750(General Electric, Fairfield connectivity, USA), equipped with rapid gradients. Foam filled head coils are used during scanning to reduce movement of the subject's head, and to provide the subject with soft earplugs, to carefully position the subject's head within the coils, and to provide comfortable support. The patient is guided before scanning, the patient is informed to keep still, the eyes are closed, the patient does not want any special thing as far as possible, and the patient is confirmed not to enter a sleep state in the scanning process after the scanning is finished. The main scan parameters are as follows:

high spatial resolution structural data (MRI) acquisition: and collecting structural image and functional image data under a Siemens 3Tesla superconducting magnetic resonance instrument. The structure was scanned using a T1-SPGR sequence with TR 8.5ms, TE 3.4ms, flip angle 12 °, FOV24 cm × 24cm, scan matrix 512 × 512, layer thickness 1mm, and total 156 slices.

Rest state functional magnetic resonance data (fMRI) acquisition: GRE-EPI sequence acquisition with TR 2000ms, TE 30ms, flip angle 90 °, FOV240 × 240mm, scan matrix 64 × 64, slice 30, layer thickness 5.0mm, voxel 3.75 × 3.75 × 5mm, 250 time points per acquisition.

Diffusion tensor Data (DTI) acquisition: single-shot spin-echo planar sequence collection: TR 9000ms, TE 79.7ms, NEX 2, FOV24 cm × 24cm, scan matrix 256 × 256, and layer thickness 3 mm. 45 layers, no interlayer spacing, 30 directions.

(2) Image data analysis

1. Structural morphology analysis: optimized VBM method (Optimized volume-based morphology, OVBM)

Creating a symmetric average whole brain template: all the original images were converted to a size of 1 × 1 × 1mm3 while being spatially smoothed to create a T1 template.

Creating an ash template: the original image is registered to the average T1 template generated in step (1), the gray matter pattern is extracted by segmentation, then spatially gaussian smoothed, and the averages are superimposed.

Creating a mirror image for each tested object.

Registering the tested original image and the created mirror image to the generated template for segmentation: and standardizing the original gray matter image by using the created gray matter template, and writing the obtained standardized parameters into the corresponding original image to obtain a final standardized whole brain image. The normalized image is then segmented and the resulting gray matter image is then smoothed.

Sixthly, performing two-sample t test based on voxel on gray matter density/volume of two groups of data of the original image and the mirror image to obtain a parameter statistical chart, and taking a region with P <0.05 as difference after multiple comparison and correction.

Seventhly, exploring the correlation between abnormal brain structures and the quality characteristics and the state characteristics of the depression.

2. And (3) constructing a functional network: correlation method based on seed points

Selecting a region of interest (ROI) based on a priori assumption or a structural image analysis result.

And extracting the average time sequence of all voxels in each ROI as a time sequence signal of the ROI.

Thirdly, time domain preprocessing is carried out on the time sequence signal of each ROI, and the preprocessing comprises the step of regressing a tested head moving signal; band-pass filtering to remove machine noise and physiological noise;

calculating correlation coefficients for the ROI time sequence between every two ROI time sequences, and normalizing all the correlation coefficients to normal distribution by using Fisher r-z transformation on the obtained correlation coefficients; each ROI is used as a node in the network, and the correlation coefficient between the two nodes is used as an edge between the nodes to obtain the whole function network related to the depression;

for the functional network, obtaining the total connectivity (functional connection strength) of each node by adopting the n-to-1 connectivity in the graph theory;

sixthly, counting and comparing the change of the node degree of the network of the brain between the patient and the normal tested function and searching the physiological mark. The correlation of the presence of abnormal brain structural connections in depression with the predisposition and status characteristics was explored.

3. And (3) constructing a functional network: whole brain function connection network construction

Segmenting an anatomical region: segmenting the pre-processed functional data using an a priori anatomical template (e.g., an anatomical auto-labeling template, AAL) to obtain 116 an anatomical region as a region of interest (ROI);

extracting the average time sequence of all voxels in each ROI as a time sequence signal of the ROI;

thirdly, time domain preprocessing is carried out on the time sequence signal of each ROI, and the preprocessing comprises the step of regressing a tested head moving signal; band-pass filtering is carried out to remove machine noise and physiological noise;

calculating correlation coefficients for the ROI time sequence between every two ROI time sequences, and normalizing all the correlation coefficients to normal distribution by using Fisher r-z transformation on the obtained correlation coefficients; obtaining a 116 × 116 correlation coefficient matrix;

establishing significance of the correlation coefficient matrix to obtain a side with statistical significance so as to obtain a functional connection matrix of patients and normal testees;

seventhly, obtaining the functional network parameters of each group of people by using indexes in graph theory analysis, wherein the indexes comprise the properties of node degree, clustering coefficient, shortest path, small world and the like;

counting the functional network parameters of the generalized anxiety patient and the normal tested brain, and searching the imaging marker of the large-scale functional network level. The correlation of the presence of abnormal functional connections in depression with the predisposition and status characteristics was explored.

4. And (3) constructing a structural network: diffusion Tensor Imaging (DTI)

Diffusion tensor imaging of the brain (for finding links in white matter) and high spatial resolution T1 weighted magnetic resonance imaging (for determining nodes in gray matter)

② distinguishing and dividing white matter and gray matter

Thirdly, the whole brain white matter nerve fiber bundle is obtained by calculation of the linear flow tracing algorithm

Four anatomical partition of gray matter and further subdivision into 116 functional areas (consistent with the definition of nodes of the fMRI functional network)

Fifthly, utilizing the node and side information provided by (3) and (4) to construct a structure connection network

Sixthly, for the structure network, the indexes in the graph theory analysis are also adopted, and the characteristics including node degree, clustering coefficient, shortest path, small world and the like are adopted to obtain the structure connection network parameters of each group of crowds

And seventhly, statistically comparing the parameters of the connection network of the patient and the normal tested brain structure, searching the imaging marker of the connection network level of the large-scale structure of the patient, and exploring the correlation between the abnormal structure connection and the quality characteristics and state characteristics of the depression.

5. Pairwise spatial clustering (SPC): building a functional network

Abnormal changes in the affective regulatory system (involving 17362 cortical voxels, and all possible 15.071 billion pairs of functional connections) in depression patients were studied relative to normal using a Pairwise spatial clustering (SPC) and Network-based analysis (NBS) approach. SPC is used to find grey matter regional pairs with abnormal functional connections for depression patients, and NBS is used to count abnormal conditions at the network level for such abnormal grey matter regional pairs. Deep exploration on neuropathological changes of a depression emotion regulation system from a voxel level to a connection level and further to a network level; firstly, a sub-network with abnormal functional connection is found, and a prediction model is constructed by adopting a Support Vector Regression (SVR) method on the basis of the sub-network, so that the quality characteristics and the state characteristics related to depression are predicted.

6. Static dynamic function network

Firstly, partitioning the gray matter by adopting an AAL template, and subdividing the gray matter into 1024 high-precision functional areas;

extracting the average time sequence of all voxels in each ROI as a time sequence signal of the ROI; performing frequency division and pretreatment on the ROI signal to obtain neuron level signals of different frequency bands;

calculating the phase difference of each time point between every two signals between the regions in the frequency domain as the dynamic function connection strength between the two regions at the time point to obtain a dynamic function connection matrix between the ROIs;

and fourthly, obtaining dynamic function network parameters of each group of people by using the characteristics of graph theory indexes (node degree, clustering coefficient, shortest path, small world and the like), constructing key nodes (brain function areas) of the brain loop of the depression patient, forming a brain dynamic Rich Club loop, and analyzing whether the dynamic characteristics of the brain key loop of the depression patient are changed or not.

7. Pattern recognition classification model for depression

For each depression patient, local indexes and connection indexes of tested brain areas of functions and structures are extracted (the brain function dynamic ALFF indexes of 90 brain areas, the brain structure VBM indexes of 90 brain areas, 4005 DTI structure connection indexes, 4005 resting state function connection indexes and about 8200 brain function structure characteristics in total);

calculating the correlation of each feature with the depression scale, and keeping the features related to the scale as disease-related features;

thirdly, the dimensionality of the remaining features is higher, and the remaining features are subjected to typical correlation analysis with a disease scale to find out main components related to the disease scale;

extracting 3 main components with the highest interpretation degree, and clustering all tested components into n types by adopting a hierarchical clustering method, wherein the value of n is determined by adopting a Silhouete criterion;

analyzing the recovery probability of each subtype to provide basis for the prediction of depression recovery.

In conclusion, by adopting the method provided by the invention, a multi-mode image analysis method can be developed, and the function of extracting the characteristic information of brain function and structural activity is realized;

Claims

1. A method for studying depression, characterized in that during the course of the study, the analysis is carried out on patients with unidirectional depression and bidirectional affective disorder in the period of depressive episodes and as a reference; in the clinical symptom evaluation process, factors combining the quality characteristic evaluation and the state characteristic evaluation are used as research indexes, and the method comprises the following implementation steps:

2. The method of claim 1, wherein the image data acquisition comprises high spatial resolution structural data acquisition, resting state functional resonance data acquisition, diffusion tensor data acquisition;

3. The method of claim 2, wherein during the structural morphology analysis, a symmetric average whole brain template is created, gray matter templates are created, mirror images are created for each test, the original image and the created mirror images are registered to the generated template for segmentation, two voxel-based sample tests are performed on gray matter density/volume of the two sets of data of the original image and the mirror images, parameter statistical maps are obtained, and areas with P <0.05 as difference after multiple comparison corrections are taken.

4. The method for studying depression according to claim 2, wherein in the functional network construction process, the region of interest is selected based on a priori assumption or structural image analysis result;

5. The method for studying depression according to claim 2, wherein a whole brain function connection network is used in the process of constructing the functional network, and specifically diffusion tensor imaging is included in the process of constructing the structural network.

6. The method of claim 2, wherein the local index and the connection index of the tested function and structure of each brain region are extracted from the pattern recognition classification model of depression, and the correlation with the depression scale is calculated for each feature.