Biology subjects

Baliki, M. N.

Publications and source records attributed to Baliki, M. N..

3 recordsLinked to original sources

bioRxiv2025

Revisiting Amplitude of Low-Frequency Fluctuations (ALFF) in Resting-state fMRI: Clarifications and Improvements

The amplitude of low-frequency fluctuations (ALFF) and its related measure, fractional ALFF (fALFF), are widely used resting-state fMRI techniques for quantifying spontaneous neural activity within specific frequency bands. However, inconsistencies in the definition and implementation of ALFF have led to confusion in the field. In this study, we provide a mathematical clarification of ALFF and fALFF by introducing two variants: the arithmetic mean-defined ALFF/fALFF (amALFF/amfALFF) and the quadratic mean-defined ALFF/fALFF (qmALFF/qmfALFF). We examine the relationships between mean BOLD intensity (MBI), amALFF, and qmALFF across both subjects and voxels using two independent datasets mapped onto different brain templates. Additionally, we investigate the impact of z-scoring the original BOLD signal on ALFF and fALFF metrics. Our key findings include: (1) MBI is positively correlated with both amALFF and qmALFF, highlighting the need for normalization to subject-level means; (2) normalized qmALFF and qmfALFF are highly correlated with normalized amALFF and amfALFF, respectively, at both the subject and voxel levels; (3) z-scoring the BOLD signal does not affect amfALFF or qmfALFF, but it substantially alters amALFF and qmALFF. Based on these findings, we present a comprehensive flowchart of the (f)ALFF algorithm implemented in the temporal domain. The full procedure is implemented in R, and the corresponding script is available at: https://github.com/lejianhuang/ALFF.

neuroscience↗

bioRxiv2022

Widespread, perception-related information in the human brain scales with levels of consciousness

How does the human brain generate coherent, subjective perceptions--transforming yellow and oblong visual sensory information into the perception of an edible banana 1? This is a hard problem. The standard viewpoint posits that anatomical and functional networks integrate local, specialized processing across the brain to somehow construct unique percepts. Here, we provide evidence for a novel organizational concept by uncovering task-specific information distributed across the human brain. First, we show that functional magnetic resonance imaging (fMRI) can uncover task-specific information throughout the neocortex, even across voxels traditionally discarded as "noise" (t-statistics {approx} 0), challenging the sensitivity of traditional linear, univariate analytical approaches. Remarkably, task-specific signals could also be uncovered from across-subject variances and were ubiquitous even in the subcortex and cerebellum. Finally, we show that the widespread signal in regions remote from a tasks primary and secondary sensory cortices depends on the level of sedation, suggesting it is related to perception{dagger} rather than sensory stimulus encoding. We hypothesize that these widespread, task-specific, and consciousness level-dependent signals may be the basis for coherent, subjective perceptions.

neuroscience↗

bioRxiv2020

The Hard Limits of Decoding Mental States: The Decodability of fMRI

High-profile studies claim to assess mental states across individuals using multi-voxel decoders of brain activity. The fixed, fine-grained, multi-voxel patterns in these "optimized" decoders are purportedly necessary for discriminating between, and accurately identifying, mental states. Here, we present compelling evidence that the efficacy of these decoders is overstated. Across a variety of tasks, decoder patterns were not necessary. Not only were "optimized decoders" spatially imprecise and 90% redundant, but they also performed similarly to simpler decoders, built from average brain activity. We distinguish decoder performance when used for discriminating between, in contrast to identifying, mental states, and show even when discrimination performance is strong, identification can be poor. Using similarity rules, we derived novel and intuitive discriminability metrics that capture 95% and 68% of discrimination performance within- and across-subjects, respectively. These findings demonstrate that current across-subject decoders remain inadequate for real-life decision making.

neuroscience↗