| مستخلص: |
Background To understand the complex mechanisms of schizophrenia, interdisciplinary approach such as complexity science has integrated the essences of physics, mathematics and computational neuroscience, becoming essential to decrypt heterogeneous brain information. Nonlinear dynamical approaches to quantify the complexity of brain signal extract fundamental features from spatial-temporal neuroimaging data at multiple levels. Power law scaling as a well-validated principle in physics is used to describe the complex nature of a system across time scales. Following "the loss of brain complexity hypothesis" (Yang & Tsai, 2013), we anticipate that neuronal dynamics in healthy states exhibits multiscale variability, a characteristic of power law behavior, and pathological state is associated with the breakdown of neuronal dynamics. In this research, we adopt the nonlinear property of a complexity, investigating the change of power-law characteristics in a large-scale schizophrenic and healthy resting-state fMRI data. Methods Brain image data of age and sex-matched 200 schizophrenia and 200 healthy subjects (age mean=43.56; male = 49.5% for each group), right-handed Han Chinese, were retrieved from Taiwan Aging and Mental Illness (TAMI) cohort. Whole brain resting-state fMRI and anatomical MRI image data are acquired to indicate the dynamic activities across brain regions. For the patient's group, the average onset is 28 years old, and the average duration of onset is 15 years. Image preprocessing was operated with DPARSF (V4.3) and SPM12. To extract power law scaling, edited Pwelch function was processed under MATLAB2017a. With the data in frequency domain presented in logarithm plot, linear regression is applied to capture the pattern of scaling. Lastly, general linear model is used to compare these slopes between 2 groups with age and sex controlled for each voxel. Results Whole brain grey matters of 55749 voxels were searched with the extent threshold k= 35 voxels (p =0.02). The expected false discovery rate is ≤0. Statistical images were assessed for cluster-wise significance using a cluster-defining threshold of uncorrected p ≤ 0.001. The results show that schizophrenia patients has significantly more positive power law spectrum slope than healthy adults at 4 clusters: left precuneus (k=17555, T=7.72) and left middle occipital gyrus (T=6.97), left medial dorsal nucleus (k =183, T=5.99), right inferior frontal gyrus (k =160, T=4.26), and right middle temporal gyrus (k =48,T=3.93). These 4 clusters have p (FRD-cor) <0.001 at voxel level. On the other hand, 2 key regions were identified where healthy group show significant higher power law slope: right putamen (k =60, T=-3.11) and left putamen (k =44, T=-3.11). Discussion The located regions with complexity abnormality found in this research indicates over or insufficient brain activities in the schizophrenic brain, corresponding with clinical observations such as auditory hallucinations, attentional control and stop-signal inhibition. While adopting complexity analysis in patient's data, the heterogeneity of mental illness and typical development of the brain should be considered. The diagnoses of patients here were conducted by board-certified psychiatrists. In addition, the frequency filter of image signal should be defined with specific bands related to physiological responses to approach actual activation information. In the near future, power law scaling has great potential to offer a potential diagnostic biomarker of psychiatric disorders. [ABSTRACT FROM AUTHOR] |
Full Text Finder