| Schizophrenia(SZ)is a chronic,severe,disabling,and highly genetic mental disorder that seriously affects patients and their families.It is a serious public health and social problem.Antipsychotic medication is the first choice for treatment of schizophrenia.However,using standard antipsychotic medications,there are still some patients who have not shown significant clinical improvement.Electroconvulsive therapy(ECT)is a safe and efficient treatment for schizophrenia,especially for patients with schizophrenia who need rapid improvement and relief of acute symptoms,patients with suicidal tendency and anxiety,and patients with treatment-resistant schizophrenia who are resistant to antipsychotic medications.ECT may be more effective in combination with medication.ECT can quickly improve symptoms,reduce symptoms,and reduce the rate of psychiatric hospitalization,but not all patients achieve significant improvement after receiving ECT.Considering the economic burden of ECT,extended treatment time and possible side effects,the predictive evaluation of clinical efficacy of ECT before treatment to determine whether the patient is suitable for ECT is of great value,which may help clinicians,patients and their relatives consider the potential benefits and costs of ECT.In addition,brain regions related to clinical improvement may help researchers better understand the underlying mechanism of ECT.Therefore,based on structural Magnetic Resonance Imaging(s MRI),Diffusion Tensor Imaging(DTI),and resting-state functional Magnetic Resonance Imaging(rs-f MRI),this study analyzed the differences in imaging patterns of schizophrenia patients with different responses to ECT combined with antipsychotics,and further explored neuroimaging markers based on magnetic resonance images that could be used to predict the efficacy of ECT in schizophrenia patients.Specifically,the main results of this dissertation include the following aspects:Firstly,we explored the baseline image differences of schizophrenia patients with different ECT responses based on statistical analysis to determine brain regions and indicators with potential predictive power.The voxel-based morphometry analysis(VBM)and tract-based spatial statistics(TBSS)methods were used to compare the differences in the structure of gray matter and white matter between the responders(RS)and non-responders(NRS)in patients with schizophrenia.We also analyzed various indicators such as amplitude of low-frequency fluctuations(ALFF),fractional amplitude of low-frequency flctuation(f ALFF),regional homogeneity(Re Ho)and degree centrality(DC)capture the functional differences of different spontaneous brain activities in the two groups.The results of the chapter showed that patients in RS group had larger gray matter volumes in the temporal gyrus,hippocampus,insula,and precuneus than in NRS group.This chapter also showed that the functional activities of the two groups in the occipital gyrus,frontal gyrus,cuneus,fusiform and other brain regions were different.In addition,based on correlation analysis,we also found a significant correlation between these brain areas and ECT response.The results provide potential biomarkers for future personalized predictions of whether patients will respond,and provide more evidence for the interpretation of the mechanism of action and response of ECT.Secondly,we continued to explore the neuroanatomical characteristics in predicting the efficacy of ECT combined with antipsychotics in schizophrenia.The structural basis is critical to the pathophysiology of schizophrenia,and several studies have confirmed the predictive ability of gray matter volume on response to ECT in depression.Our previous work also obtained the differences in gray matter volume from schizophrenic patients with different responses to ECT.Therefore,this chapter was based on the gray matter volume in difference area,extracting first-order statistical radiomics features,and constructing a classification model to predict whether patients with schizophrenia will respond to ECT.The results of this chapter showed that the neuroanatomical features based on structural images can distinguish between RS and NRS,with an accuracy of 90.91%.The classification performance of structural features was further confirmed in the test set,and the accuracy was 87.59%.In addition,important neuroanatomical features with predictive power included the cortex(inferior frontal gyrus,cingulate cortex,temporal lobe and parietal lobe)and subcortical areas(insula,thalamus and hippocampus).Useful neuroanatomical features for prediction involved areas of the brain that are regulated by ECT.Thirdly,we completed the study of predicting the efficacy of ECT combined with antipsychotics in schizophrenia based on the gray matter structure characteristics and white matter connection characteristics in the strong electric field area.In ECT,a strong current is applied directly to the brain through electrodes located near the temporal or frontal lobes.The areas of the brain with strong electric field distribution are usually most affected by ECT.Therefore,the characteristics based on these area with the strongest charge density may have better predictive power in predicting ECT efficacy.The neuroanatomical features useful for prediction in the previous work mostly involved the brain regions regulated by ECT,especially the gray matter features of the frontotemporal region.The white matter integrity of schizophrenia is abnormal,especially the frontotemporal connection.Therefore,we took the strong electric field areas as the regions of interest(ROIs),extracted the statistical features based on the gray matter volume of ROIs,and then extracted the statistical features based on the white matter connections between ROIs,and constructed regression prediction models for single-type features and fusion features.The results of the chapter showed that the radiomics features based on multi-parameter MRI had the potential to predict the response of a patient to ECT.The regression model had a low root mean square error of 14.980 on the test set,and the Pearson correlation coefficient between the predicted value and the actual value of 0.777.In addition,the prediction results also showed that the regression model based on the fusion features of white matter and gray matter had better prediction performance than the model constructed with only gray matter or white matter features.The gray matter characteristics and white matter characteristics based on the strong electric field region may serve as potential predictors of response to ECT in schizophrenia.Then,we continued to complete the study of predicting the efficacy of ECT combined with antipsychotics in schizophrenia based on dynamic functional network characteristics.Functional magnetic resonance imaging measures changes in the hemodynamics of the brain to determine neural activity.Therefore,it could be used to measure brain function and should be considered when evaluating the effect of ECT,especially the dynamic characteristics of the brain function network(temporal variability).Therefore,we not only investigated the difference in temporal variability between the RS and NRS,but also further determined the relationship between temporal variability and symptom improvement and the predictive performance of temporal variability on ECT response in patients with schizophrenia.The results of the chapter showed that,compared with NRS,the temporal variability of the triangular part of inferior frontal gyrus(IFGtriang.R)in RS was significantly lower,while the temporal variability of the left temporal pole(TPOsup.L)and right middle temporal gyrus(MTG.R)in RS was significantly higher.In the combined patient group,the degree of remission was negatively correlated with the temporal variability of IFGtriang.R,and positively correlated with the temporal variability of TPOsup.L and MTG.R.The results of classification analysis showed that the temporal variability had good performance in distinguishing RS and NRS,and the area under the curve is 0.8969.In regression analysis,support vector regression and linear regression models also showed good predictive performance,with root mean square errors of 16.58 and 15.36,respectively,and the Pearson correlation coefficients between the predicted and actual values were 0.67 and 0.72,respectively.The results showed that the temporal variability could predict the efficacy of ECT in patients with schizophrenia,and might help explain the improvement of symptoms from the perspective of the instability of the intrinsic functional connection in the resting state.Finally,we completed the study on the prediction of ECT combined with antipsychotic in schizophrenia based on multimodal magnetic resonance imaging.Multi-parameter MRI is not only beneficial to enhance the performance of predictive model,but also may reveal key information about the pathophysiology of schizophrenia.This chapter explored whether a multi-parameter image ensemble model based on the integration of single image classification models such as s MRI,DTI,and rs-f MRI could provide better prediction performance.In addition,we also integrated clinical features and imaging features to build a multi-modal model for prediction the response to ECT,and explored the contribution of clinical features to prediction enhancement.The results showed that the model constructed by multi-parameter image fusion features including a variety of magnetic resonance images had a more accurate predictive ability(accuracy rate of 93.39%)than single-sequence images(accuracy rate ranges from 75.19% to 87.84%),but no significant improvement in the prediction of clinical features was found.This research provides evidence that multi-modal neuroimaging features could provide more comprehensive and richer information and have more predictive potential in the prediction of ECT efficacy.In summary,based on the gray matter volume of the s MRI,the diffusion characteristics of DTI,the region-based functional characteristics,the connection-based connection characteristics and the temporal variability of rs-f MRI,we investigated the difference in imaging patterns of schizophrenia patients with different ECT responses.Based on single-sequence images or multi-parameter images or local specific regions or whole brain regions,we continued to gradually explore and dig deeper around the predictive ability of ECT response in schizophrenia.This study proves from the imaging perspective that the neuroimaging features based on MRI might be a potential predictor of response to ECT in schizophrenia,which provides further evidence for the application of neuroimaging to ECT response prediction.In addition,this study might help explain the underlying mechanism of ECT and help improve personalized clinical treatment in the future. |
PDF Full Text Request