Selection of individualized TMS targets and prediction of efficacy for schizophrenia

Abstract

The present invention discloses a method for selecting individualized targets for TMS intervention in treating schizophrenia patients and predicting the treatment effect, comprising the following steps: step 1, collecting patient head magnetic resonance scanning data before and after treatment and PANSS scores before and after treatment; step 2, obtaining the change in PANSS scores of patients before and after intervention; step 3, preprocessing the resting state fMRI; step 3, epicenter analysis at the group level; step 4, determination of individual 'epicenters'; step 5, prediction of rTMS intervention targets; step 6, performing spearman correlation analysis on the change in PANSS scores before and after intervention and the deviation distance. The present invention combines the epicenter with functional connectivity, and provides a new individualized target selection method for rTMS precise treatment of schizophrenia patients.

Description

Technical Field

[0001] The present invention belongs to the field of image processing technology, and in particular relates to a method for selecting individualized treatment targets for schizophrenia patients through rTMS intervention and predicting treatment effects. Background Art

[0002] Schizophrenia (SZ) is a serious mental disorder with the main symptoms of positive symptoms, negative symptoms, cognitive impairment, etc. The disease usually occurs in late adolescence or early adulthood, with significant disability and morbidity, and brings a serious economic burden to patients and their families. At present, the treatment of schizophrenia is mainly antipsychotic drugs, but for one-third of people, this treatment is ineffective and eventually develops into refractory schizophrenia. Therefore, for patients with refractory schizophrenia, it is urgent to develop treatments other than drugs. Non-invasive brain stimulation therapy provides a treatment for schizophrenia other than drugs. Transcranial Magnetic Stimulation (TMS) was originally developed as a method to treat patients with depression. Currently, the US Food and Drug Administration has approved TMS as an add-on therapy for routine treatment of depression and obsessive-compulsive disorder, and the popularity of TMS has increased significantly. Because TMS has extremely low risks, adding TMS to routine treatment with antipsychotic drugs will not increase side effects, but may increase the chances of successfully treating symptoms of mental illness.

[0003] Generally speaking, TMS is applied in two different ways: single pulse and repetitive pulse. Single pulse TMS is used in more experimental settings and to determine motor thresholds, whereas repetitive pulse TMS (rTMS) is used to treat a variety of disorders, including major depressive disorder, obsessive-compulsive disorder, and schizophrenia. The beneficial effects of rTMS on psychiatric disorders can be explained by short-term modulation of neuronal communication and subsequent changes in neuronal plasticity, as evidenced by changes in resting cerebral blood flow after rTMS. The neuromodulatory effects of TMS also depend on the frequency applied, the choice of the interstimulus interval, and the choice of target.

[0004] There are currently a large number of studies showing that TMS can effectively relieve the symptoms of schizophrenia. However, many studies have not found the same results, and the results are heterogeneous. The reason for the heterogeneity may be due to the heterogeneity between individuals. Even if schizophrenia patients have similar functional impairments, their symptoms vary greatly. Therefore, current research hopes to find some individualized indicators to treat patients, such as individualized targets and individualized stimulation frequencies. Previous studies have calculated the maximum anti-correlation point of the connection between sgACC and DLPFC as an individualized target. The results showed that the farther the distance between the individual target and the actual target, the worse the effect.

[0005] In the past decade of research, researchers have been seeking to find neuroimaging features of schizophrenia that can be used for diagnosis, prediction of the course of the disease, and expected treatment response. However, due to the high heterogeneity of schizophrenia patients, progress is currently slow and the results are also heterogeneous. Therefore, recent studies have proposed an "epicenter" model, in which the disease process begins at the epicenter and then spreads lesions to other areas of the brain through potential connectivity patterns. This connectivity-based epicenter model has been successfully found to reveal the onset of disease in local areas of the brain and to a certain extent predict the neurophysiological development of neurodegenerative diseases. Since schizophrenia is generally considered to be a disorder of functional networks or a disorder of disconnection, a reduction in gray matter volume and thinning of cortical thickness are commonly found in schizophrenia patients. Based on the combination of gray matter volume changes at the patient group level and the potential connectivity pattern, the epicenter at the group level is obtained. In schizophrenia, some individual differences in abnormal patterns of neuroanatomy may be due to differences in individual origins, so it is necessary to explore the differences between the epicenter at the individual level and the epicenter at the group level. Summary of the invention

[0006] The purpose of the present invention is to overcome the shortcomings of the prior art and provide a method for selecting individualized TMS targets for schizophrenia and predicting the efficacy. The present invention combines the potential functional connectivity with the gray matter volume atrophy to obtain the epicenter, combines the group level epicenter with the individual level epicenter, and selects the five strongest epicenters as the predicted epicenters, and finally obtains the rTMS individualized optimal target. The influence of the distance between the traditional target and the individual optimal target on the therapeutic efficacy of rTMS is studied, which provides a new perspective for rTMS treatment of schizophrenia patients and provides a method for finding individual targets.

[0007] The object of the present invention is achieved through the following technical solution: a method for selecting individualized TMS targets for schizophrenia and predicting the therapeutic effect, comprising the following steps:

[0008] Step 1: Perform rTMS intervention treatment with traditional targets on patients with schizophrenia, collect the patients' head MRI scan data before and after treatment and PANSS scores before and after treatment, and the MRI scan data include resting state fMRI and structural MRI data;

[0009] Step 2: Subtract the PANSS score before treatment from the PANSS score after treatment to obtain the change in the patient's PANSS score before and after intervention;

[0010] Step 3: Preprocess the resting-state fMRI. The preprocessing steps include: ① removing the data of the first 5 time points; ② performing time point correction; ③ correcting for head movement; ④ using structural images to generate a DARTEL template and perform standardization, and resampling to 3 mm; ⑤ smoothing using an 8 mm Gaussian smoothing kernel; ⑥ regressing white matter, cerebrospinal fluid, and linear trends; ⑦ filtering;

[0011] Step 4: Epicenter analysis at the group level: Download the functional MRI data and structural MRI data of the schizophrenia patient group from the public data set, and download the functional MRI data and structural MRI data of the healthy subject group from the Human Connectomics Project; First, use the T1 structural data in the MRI data of the two groups of subjects to obtain gray matter, white matter and cerebrospinal fluid images through voxel-based morphometric measurement; Use the gray matter image to calculate the gray matter volume of the gray matter cortex, and then adjust the GMV of each voxel by regressing the effects of gender, age, age squared, total intracranial volume and location using a regression model; Subsequently, use a gray matter mask to exclude non-gray matter voxels; For each person with schizophrenia, use z-score normalization to normalize the adjusted GMV value to obtain an individualized z-score map of gray matter reduction voxels;

[0012] A set of standardized functional connection groups were derived from the functional magnetic resonance imaging data of healthy subjects: the whole brain partitions were composed of 200 cortical and 54 subcortical regions, each region was used as a separate seed point, the functional connection between each seed point of the healthy subject and the rest of the seed points in the whole brain was calculated, and the FC value of each seed point was mapped back to the whole brain atlas, and 254 FC maps were obtained for each subject; then the FC map was converted into an FC-z map by Fisher-z transformation; for the FC-z map obtained for each seed point, a single-sample t-test was performed on the FC-z map at the subject level to obtain a statistical FC-t map at the group level;

[0013] These 254 regional FC-t maps were then used to detect the center of gray matter atrophy in each patient. The voxel-individualized z-score maps were first averaged at the seed point level to obtain 254 values, and then the 254 values ​​were mapped back to the whole-brain atlas to obtain the gray matter reduction z-score map. The gray matter reduction z-score map was spatially correlated with the FC-t map of each seed point of the healthy subjects, and the Spearman correlation coefficient was calculated and further converted to Fisher-z value to quantify the GOF score. The GOF score was subjected to a one-sample t-test at the group level to determine the significance of the candidate epicenter. The area with P < 0.001 in the one-sample t-map was the epicenter area at the group level.

[0014] Step 5. Determination of individual epicenters: Use the resting-state fMRI data before rTMS intervention after preprocessing in step 3 and the resting-state fMRI data of the healthy control group collected at the same location; process all fMRI data to obtain the gray matter atrophy z-score map of each person with schizophrenia, average the values ​​of the gray matter atrophy z-score map on 254 whole-brain partition templates, and obtain a gray matter atrophy z-score matrix with a column length of 254; then calculate the functional connectivity between the seed points of the healthy subjects, and calculate the functional connectivity between two seed points for a total of 254×254 times to obtain a 254×254 FC matrix; for each seed point, use Spearman correlation analysis to calculate the correlation between the FC matrix and the gray matter atrophy z-score matrix, and the obtained correlation coefficient is recorded as r; at the same time, use a non-parametric test to correct the P value obtained in the Spearman correlation analysis calculation to obtain P spin ; P spin Areas with a positive r value < 0.05 were defined as individual epicenters;

[0015] Step 6. Prediction of rTMS intervention targets: 2% of the area was used as the target prediction epicenter. First, the epicenter at the group level and the individual epicenter were intersected, and the first 2% of the area was taken as the prediction epicenter. If the intersection was less than 2%, the individual epicenter was supplemented to 2%; then, the resting state data obtained from the pre-intervention scan was used after preprocessing in step 3 to calculate the functional connections of all voxels in the left dorsolateral prefrontal cortex with the predicted epicenter; each subject found five points with the strongest functional connection with the epicenter, and then found a point in the left dorsolateral prefrontal cortex with the closest Euclidean distance to the five points, and this point was considered to be the best intervention target;

[0016] Step 7: Calculate the Euclidean distance between the optimal target and the traditional stimulation target for each subject. This distance is defined as the deviation distance. Perform a spearman correlation analysis on the change in PANSS score before and after the intervention and the deviation distance. The result is recorded as p. If p < 0.05 and r < 0, it is considered that there is a significant negative correlation between the deviation distance from the optimal target and the score change.

[0017] The beneficial effect of the present invention is that the present invention combines the epicenter with the functional connection, and provides a new method for selecting individualized targets for rTMS precise treatment of schizophrenia patients. It is also found that the closer the distance between the actual target and the individualized intervention target, the better the rTMS intervention effect and the more symptom improvement, which reveals the contribution of the epicenter to the study of individualized targets and provides a new perspective for the subsequent establishment of precise targets. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 This is a flow chart of the method for selecting individualized TMS targets and predicting therapeutic efficacy for schizophrenia of the present invention;

[0019] Figure 2 To predict the correlation between the distance between the optimal target and the traditional target and the positive scale reduction of the PANSS scale. DETAILED DESCRIPTION

[0020] The technical solution of the present invention is further described below in conjunction with the accompanying drawings.

[0021] like Figure 1 As shown, a method for selecting individualized TMS targets and predicting therapeutic effects for schizophrenia of the present invention comprises the following steps:

[0022] Step 1: Perform rTMS intervention treatment with traditional targets on patients with schizophrenia, collect the patients' head MRI scan data before and after treatment and PANSS scores before and after treatment, and the MRI scan data include resting state fMRI and structural MRI data;

[0023] Step 2: Subtract the PANSS score before treatment from the PANSS score after treatment to obtain the change in the patient's PANSS score before and after intervention;

[0024] Step 3: Preprocess the resting-state fMRI. The preprocessing steps include: ① removing the data of the first 5 time points; ② performing time point correction; ③ correcting for head movement; ④ using structural images to generate a DARTEL template and perform standardization, and resampling to 3 mm; ⑤ smoothing using an 8 mm Gaussian smoothing kernel; ⑥ regressing white matter, cerebrospinal fluid, and linear trends; ⑦ filtering;

[0025] Step 4. Epicenter analysis at the group level: Download the functional MRI data and structural MRI data of the schizophrenia patient group from the public data set, and download the functional MRI data and structural MRI data of the healthy subject group from the Human Connectome Project (HCP). First, the T1 structural data in the MRI data of the two groups of subjects were used to obtain gray matter, white matter and cerebrospinal fluid images through voxel-based morphometry (VBM). Only gray matter images were used in the present invention. The gray matter volume of the gray matter cortex was calculated using the gray matter image, and then the GMV of each voxel was adjusted by using a regression model to regress the effects of gender, age, age squared, total intracranial volume and location; subsequently, a gray matter mask was used to exclude non-gray matter voxels; for each person with schizophrenia, the z-score (GMV 精分 -meanGMV 健康 ) / stdGMV 健康 ) The adjusted GMV values ​​were normalized relative to those of healthy subjects to obtain individualized z-score maps of gray matter reduction voxels.

[0026] It was assumed that the intrinsic connectivity pattern of the epicenter was most similar to the atrophy pattern of the patient. To determine the potential disease epicenter of neuroanatomical abnormalities, first, a set of normative functional connectivity (FC) groups were derived from the functional magnetic resonance imaging data of healthy subjects in HCP: a whole-brain partition consisting of 200 cortical regions (Schaefer200 Atlas) and 54 subcortical regions (Melbourne Subcortex Atlas), each region was used as a separate seed point, and the functional connectivity between each seed point of the healthy subject and the rest of the seed points in the whole brain was calculated. The FC value of each seed point was mapped back to the whole-brain atlas, and 254 FC maps were obtained for each subject; then the FC maps were converted to FC-z maps by Fisher-z transformation. For the FC-z maps obtained for each seed point, a one-sample t-test was performed on the FC-z maps at the subject level to obtain a statistical FC-t map at the group level. These 254 regional FC-t maps were then used to detect the center of gray matter atrophy in each patient. The region with a more correlated FC map with gray matter atrophy had a higher probability of being the epicenter.

[0027] To characterize the epicenter of the patients' gray matter loss z-score maps, the voxel-individualized z-score maps were first averaged at the seed point level to obtain 254 values, and then the 254 values ​​were mapped back to the whole brain map to obtain the gray matter loss z-score map. The gray matter loss z-score map was spatially correlated with the FC-t map of each seed point of the healthy subjects, and the Spearman correlation coefficient was calculated and further converted to Fisher-z value to quantify the goodness of fit (GOF) score of the epicenter. 254 GOF scores were generated for each patient. Regions with higher GOF scores represent a higher probability of becoming candidate epicenters. Finally, a one-sample t-test was performed on the GOF scores at the group level to determine the significance of the candidate epicenters. The one-sample t map represents the inferred epicenters of the 254 regions of the whole brain at the group level; the regions with P < 0.001 in the one-sample t map of GOF are the epicenter regions at the group level; a total of 46 regions were included.

[0028] Step 5. Determination of individual epicenters: Use the resting-state fMRI before rTMS intervention after preprocessing in step 3, and the resting-state fMRI of the healthy control group collected at the same location; process all fMRI data to obtain the gray matter atrophy z-score map of each person with schizophrenia, and average the values ​​of the gray matter atrophy z-score map onto 254 whole-brain partition templates to obtain a gray matter atrophy z-score matrix with a column length of 254. Then calculate the functional connections between the seed points of healthy subjects, and calculate the functional connections between two seed points for a total of 254×254 times to obtain a 254×254 FC matrix. For each seed point, use Spearman correlation analysis to calculate the correlation between the FC matrix and the gray matter atrophy z-score matrix, and the obtained correlation coefficient is recorded as r; at the same time, use a non-parametric test to correct the P value obtained in the Spearman correlation analysis calculation to obtain P spin ; P spin Areas with a positive r value < 0.05 were defined as individual epicenters;

[0029] Step 6. Prediction of rTMS intervention targets: 2% of the area was used as the target prediction epicenter. First, the epicenter at the group level and the individual epicenter were intersected, and the first 2% of the area was taken as the prediction epicenter. If the intersection was less than 2%, the individual epicenter was supplemented to 2%; then, the resting state data obtained from the pre-intervention scan was used after preprocessing in step 3 to calculate the functional connections of all voxels in the left dorsolateral prefrontal cortex with the predicted epicenter; each subject found five points with the strongest functional connection with the epicenter, and then found a point in the left dorsolateral prefrontal cortex with the closest Euclidean distance to the five points, and this point was considered to be the best intervention target;

[0030] Step 7: Calculate the Euclidean distance between the optimal target and the traditional stimulation target (the target of this rTMS intervention for schizophrenia) for each subject. This distance is defined as the deviation distance. Perform a spearman correlation analysis on the change in PANSS score before and after the intervention and the deviation distance. The result is recorded as p. If p < 0.05 and r < 0, it is considered that there is a significant negative correlation between the deviation distance from the optimal target and the score change, that is, the closer to the optimal target, the better the therapeutic effect.

[0031] In this example, when calculating the epicenter at the population level, the data used included Shanghai Mental Health Center, the First Affiliated Hospital of Zhengzhou University, Taipei Veterans General Hospital, Chengdu Brain Science Institute, and 5 public data sets. After data quality control, a total of 1,124 schizophrenia patients and 1,046 healthy controls were included, and these subjects were used to calculate the atrophy pattern of schizophrenia patients at the group level. And when calculating the FC-z map of healthy subjects, 1,089 healthy subjects in the HCP data were used.

[0032] When calculating the individual level epicenter, the patients who underwent MRI before and after rTMS intervention were all from the Fourth People's Hospital of Chengdu, a total of 26 inpatients with schizophrenia. All patients were diagnosed with schizophrenia by professional clinical psychiatrists using SCID-I / P (Structured Clinical Interview for DSM-IV-TR, Patient Version). The Positive and Negative Syndrome Rating Scale (PANSS) was used to assess the severity of psychiatric symptoms. The total PANSS score of all patients was above 60 points, and no patient showed severe aggression or agitation. All patients received antipsychotic medication, and the daily dose of antipsychotic medication was converted to chlorpromazine equivalents (mg / d). The data of the healthy control group were collected on the same MRI scanner. In order to exclude the influence of different MRI machines, a total of 393 subjects were used.

[0033] Then, data analysis was performed according to the method of the present invention, and the experimental results were as follows: 46 areas were selected as epicenters at the group level, most of which belonged to the default mode network, executive control network and attention network. The epicenters at the individual level were heterogeneous in each individual, and it was found that the distance between the individualized rTMS intervention target and the actual rTMS intervention target was significantly correlated with the change in the positive scale score of the PANSS scale (r = -0.39, p = 0.04, as shown in Figure 2). Figure 2 As shown, the horizontal axis is the Euclidean distance between the individualized intervention target and the actual intervention target, and the vertical axis is the change in the positive scale of the PANSS scale). Correlation analysis proves that the individualized intervention target predicted by the epicenter is helpful for the efficacy, and the closer the distance, the better the effect.

[0034] The experimental results show that the effectiveness of rTMS intervention in improving symptoms of schizophrenia may depend on whether the location of the target is the best individualized target. The closer the actual target is to the individualized intervention target, the better the rTMS intervention effect and the more symptoms are improved, which reveals the contribution of the epicenter to the research on individualized targets and provides a new perspective for the establishment of precise targets in the future.

[0035] Those skilled in the art will appreciate that the embodiments described herein are intended to help readers understand the principles of the present invention, and should be understood that the protection scope of the present invention is not limited to such specific statements and embodiments. Those skilled in the art can make various other specific variations and combinations that do not deviate from the essence of the present invention based on the technical revelations disclosed by the present invention, and these variations and combinations are still within the protection scope of the present invention.

Claims

1. A method for selecting individualized TMS targets for schizophrenia and predicting the therapeutic effect, characterized in that: The following steps are involved: Step 1: Perform rTMS intervention treatment with traditional targets on patients with schizophrenia, collect the patients' head MRI scan data before and after treatment and PANSS scores before and after treatment, and the MRI scan data include resting state fMRI and structural MRI data; Step 2: Subtract the PANSS score before treatment from the PANSS score after treatment to obtain the change in the patient's PANSS score before and after intervention; Step 3: Preprocess the resting-state fMRI. The preprocessing steps include: ① removing the data of the first 5 time points; ② performing time point correction; ③ correcting for head movement; ④ using structural images to generate a DARTEL template and perform standardization, and resampling to 3 mm; ⑤ smoothing using an 8 mm Gaussian smoothing kernel; ⑥ regressing white matter, cerebrospinal fluid, and linear trends; ⑦ filtering; Step 4: Epicenter analysis at the group level: Download the functional MRI data and structural MRI data of the schizophrenia patient group from the public data set, and download the functional MRI data and structural MRI data of the healthy subject group from the Human Connectomics Project; First, use the T1 structural data in the MRI data of the two groups of subjects to obtain gray matter, white matter and cerebrospinal fluid images through voxel-based morphometric measurement; Use the gray matter image to calculate the gray matter volume of the gray matter cortex, and then adjust the GMV of each voxel by regressing the effects of gender, age, age squared, total intracranial volume and location using a regression model; Subsequently, use a gray matter mask to exclude non-gray matter voxels; For each person with schizophrenia, use z-score normalization to normalize the adjusted GMV value to obtain an individualized z-score map of gray matter reduction voxels; A set of standardized functional connection groups were derived from the functional magnetic resonance imaging data of healthy subjects: the whole brain partitions were composed of 200 cortical and 54 subcortical regions, each region was used as a separate seed point, the functional connection between each seed point of the healthy subject and the rest of the seed points in the whole brain was calculated, and the FC value of each seed point was mapped back to the whole brain atlas, and 254 FC maps were obtained for each subject; then the FC map was converted into an FC-z map by Fisher-z transformation; for the FC-z map obtained for each seed point, a single-sample t-test was performed on the FC-z map at the subject level to obtain a statistical FC-t map at the group level; These 254 regional FC-t maps were then used to detect the center of gray matter atrophy in each patient. The voxel-individualized z-score maps were first averaged at the seed point level to obtain 254 values, and then the 254 values ​​were mapped back to the whole-brain atlas to obtain the gray matter reduction z-score map. The gray matter reduction z-score map was spatially correlated with the FC-t map of each seed point of the healthy subjects, and the Spearman correlation coefficient was calculated and further converted to Fisher-z value to quantify the GOF score. The GOF score was subjected to a one-sample t-test at the group level to determine the significance of the candidate epicenter. The area with P < 0.001 in the one-sample t-map was the epicenter area at the group level. Step 5. Determination of individual epicenters: Use the resting-state fMRI data before rTMS intervention after preprocessing in step 3 and the resting-state fMRI data of the healthy control group collected at the same location; process all fMRI data to obtain the gray matter atrophy z-score map of each person with schizophrenia, average the values ​​of the gray matter atrophy z-score map on 254 whole-brain partition templates, and obtain a gray matter atrophy z-score matrix with a column length of 254; then calculate the functional connectivity between the seed points of the healthy subjects, and calculate the functional connectivity between two seed points for a total of 254×254 times to obtain a 254×254 FC matrix; for each seed point, use Spearman correlation analysis to calculate the correlation between the FC matrix and the gray matter atrophy z-score matrix, and the obtained correlation coefficient is recorded as r; at the same time, use a non-parametric test to correct the P value obtained in the Spearman correlation analysis calculation to obtain P spin ; P spin Areas with a positive r value < 0.05 were defined as individual epicenters; Step 6. Prediction of rTMS intervention targets: 2% of the area was used as the target prediction epicenter. First, the epicenter at the group level and the individual epicenter were intersected, and the first 2% of the area was taken as the prediction epicenter. If the intersection was less than 2%, the individual epicenter was supplemented to 2%; then, the resting state data obtained from the pre-intervention scan was used after preprocessing in step 3 to calculate the functional connections of all voxels in the left dorsolateral prefrontal cortex with the predicted epicenter; each subject found five points with the strongest functional connection with the epicenter, and then found a point in the left dorsolateral prefrontal cortex with the closest Euclidean distance to the five points, and this point was considered to be the best intervention target; Step 7: Calculate the Euclidean distance between the optimal target and the traditional stimulation target for each subject. This distance is defined as the deviation distance. Perform a spearman correlation analysis on the change in PANSS score before and after the intervention and the deviation distance. The result is recorded as p. If p < 0.05 and r < 0, it is considered that there is a significant negative correlation between the deviation distance from the optimal target and the score change.

Images