Multiscale Entropy of Resting-State fMRI Signals Reveals Differences in Brain Complexity in Autism

BackgroundAtypical intrinsic brain activity has been widely observed in autism spectrum disorder (ASD), yet the temporal complexity of these neural signals remains underexplored. This study aimed to characterise differences in resting-state brain signal complexity between individuals with ASD and neurotypical controls using multiscale entropy (MSE). MethodsResting-state fMRI data were obtained from the Autism Brain Imaging Data Exchange I (ABIDE I), a large multi-site dataset including 397 participants (179 with ASC, 218 neurotypical controls; ASD: Mean age 16.43 years old, SD 7.17 years old; CON: Mean age 15.75 years old, SD 5.67 years old). Voxel-wise multiscale entropy (MSE) features were extracted across multiple temporal scales. Group comparisons were conducted using voxel-wise t-tests and mixed-effects models to identify region- and scale-specific alterations in brain signal complexity. ResultsIndividuals with ASD showed reduced MSE in prefrontal regions at coarser time scales and elevated MSE in posterior midline regions, including the posterior cingulate cortex and precuneus, at finer scales, followed by a decline across coarser scales. This pattern suggests a shift toward uncorrelated randomness in posterior regions and reduced long-range complexity in frontal areas. No significant associations were found between MSE features and ADOS scores. ConclusionsThese findings reveal spatially and temporally specific alterations in brain signal complexity in ASD, particularly within the default mode network. Multiscale entropy provides a complementary approach to traditional connectivity and single-scale entropy analyses, offering novel insights into the organisation of intrinsic brain activity in neurodevelopmental conditions.