A Preliminary Resting-State Fmri Study of Regionally Reduced Complexity Patterns in Alzheimer's Disease

We have recently developed an information theoretic approach to capture stochastic patterns in BOLD fluctuations of resting-state fMRI as a mean to characterize the complexity of brain function. As a metric of complexity, the approach provides transient information (TI), which quantifies the difficulty in recognizing stochastic BOLD patterns. Complex stochastic patterns (i.e., those which carry large amounts of information) will therefore exhibit relatively high TI values, whereas random or periodic patterns carrying little information will have low TI values. We postulated that neurodegeneration causes reduction in complexity of brain function, as reflected in reduced TI values of stochastic BOLD patterns. We report preliminary findings of reduced complexity in Alzheimer's disease (AD) relative to normal brain function. Rs-fMRI data of ADNI from 11 normal (CN) subjects (mean age±std: 75±5 yrs-old) and 11 Alzheimer's disease (AD) subjects (mean age±std: 76±8 yrs-old), were analyzed. The corresponding high resolution T1 images were used for anatomical labeling of rs-fMRI using Freesurfer. TI was calculated in 90 anatomical regions, including the hippocampus, posterior cingulate, and precuneus, which are known to be affected in AD. TI values were compared between AD and CN groups using student-t tests. Table 1 lists TI values in prominent brain regions by group. AD patients had significantly reduced TI values of the hippocampus, posterior cingulate and precuneus (46%, p =0.004, 25%, p =0.01 and 23%, p =0.02, respectively) compared to CN subjects. In contrast, TI values of the motor cortex, a region largely immune to AD, were normal (p =0.1). Figure 1 shows surface rendered TI maps of a representative CN subject (male, 75 years old) and AD patient (male, 83 years old). Warm colors indicate high complexity, i.e. stochastic BOLD patterns carrying relatively high amounts of information. The findings support the idea that neurodegeneration reduces complexity of brain function. Furthermore, our findings in AD, showing reduced complexity in vulnerable brain regions, suggest that variations in complexity are potentially disease specific. Therefore, complexity of BOLD fluctuations could provide a new imaging marker for detecting functional disturbances in AD and for assessments of disease modifying interventions. Surface rendered TI maps of a representative CN subject (male, 75 years old) and AD patient (male, 83 years old). Warm colors indicate high complexity, i.e. stochastic BOLD patterns carrying relatively high amounts of information.