Mapping Neural Disintegration: DMT Analysis via EEG-fMRI Pipelines

To systematically evaluate Default Mode Network (DMN) disintegration and alpha wave collapse during the acute, immersive states induced by N,N-dimethyltryptamine (DMT), researchers deploy a highly sophisticated simultaneous electroencephalography and functional magnetic resonance imaging (EEG-fMRI) recording paradigm 1-3. This multimodal approach overcomes each imaging modality's inherent physical limitations, combining the high spatial resolution of fMRI (to map deep cortical reconfigurations) with the millisecond-range temporal resolution of EEG (to verify true, real-time neuronal oscillations) 1, 3.

The rigorous analytical workflows and computational pipelines utilized in our current research project to evaluate these two hallmark neural signatures are outlined below:

1. Methodologies for Evaluating DMN Disintegration

The Default Mode Network resides at the top end of the brain’s macroscale cortical hierarchy 4, 5. Under DMT, its cohesive internal communication collapses while its external cross-talk with other systems surges—a dual process termed "network disintegration and desegregation" 5, 6. Researchers evaluate these shifts using several distinct fMRI pipelines:

Spatial Preprocessing and Motion Gating

fMRI functional connectivity metrics are highly sensitive to head-motion artifacts 7. To ensure statistical validity, researchers implement strict motion-gating pipelines 8:

Volunteers are scanned in eyes-closed, resting-state conditions inside a 3T scanner with an eye mask to prevent accidental sensory confounding 3, 9.

Denoising steps include despiking, slice time correction, and rigid-body motion registration 8.

Scrubbing: Volumes displaying a framewise displacement (FD) exceeding a threshold of 0.4 mm are removed and replaced via interpolation 8.

To completely rule out motion as a driver of connectivity differences, researchers replicate the entire static connectivity analysis in highly matched control subsamples using a stringent threshold of FD = 0.2 mm 8.

Static resting-state Functional Connectivity (sRSFC) & Dual Regression

To measure DMN functional integrity over the peak period of the drug's effect (typically an 8-minute window post-injection), researchers employ FSL's dual regression workflow 10, 11:

Independent Component Analysis (ICA): Spatially parcellates the brain’s BOLD signals into canonical resting-state networks (RSNs), including the DMN and Frontoparietal (FP) networks 10, 11.

Step 1 Spatial Regression: The spatial components of these networks are used as spatial regressors in a General Linear Model (GLM) against each subject's 4D BOLD dataset, extracting a representative time series for each network component 11.

Step 2 Temporal Regression: These network time-series are then regressed back into the original 4D datasets, generating subject-specific spatial maps of Parameter Estimate (PE) values 11.

Integrity Quantitation: The mean PE value across voxels within the DMN is computed 11. A statistically significant reduction in DMN PE values under DMT compared to placebo indicates a loss of within-network functional integrity, physically mapping DMN disintegration 10, 11.

Between-Network Segregation Analysis

To evaluate desegregation (how much DMN boundaries dissolve to communicate with other networks), researchers compute a 7 × 7 connectivity matrix 12:

The time series extracted from Step 1 of dual regression for each network pair are entered into a mutual GLM where each RSN serves alternately as a dependent and independent variable 12.

The resulting PE values representing inter-network coupling are mapped 12. Under DMT, a significant increase in functional connectivity between the DMN, FP, and Salience (SAL) networks reveals a marked decrease in segregation 10.

Dynamic Functional Connectivity (dRSFC) and GFC

Because bolus intravenous DMT is highly dynamic, peaking at ~3 minutes and resolving by 15 minutes, static averages can hide transient network reconfigurations 13. Researchers resolve these temporal dynamics using:

Global Functional Connectivity (GFC): Measures the average Fisher Z-score correlation of a region with every other node in the brain, parcellated using the Schaefer atlas (100 regions) and AAL atlas (12 subcortical regions) 14, 15.

Sliding Window Analysis: Pairwise correlation matrices are calculated using a tapered sliding window of 44 seconds (22 TRs) convolved with a 6-second Gaussian kernel, moving in 2-second steps 16.

These dynamic GFC values are subsequently mapped to real-time, verbally reported subjective intensity ratings (convolved with a matching 6-second Gaussian filter) using linear mixed-effects (LME) models 16. This reveals that DMN GFC increases correlate linearly with the felt intensity of the trip, peaking during the initial minutes (~1 to 6 min) post-injection 16, 17.

2. Methodologies for Evaluating Alpha Wave Collapse

In the normal waking brain, parietal alpha waves (8–13 Hz) act as a vital inhibitory gate, filtering incoming sensory inputs and reflecting the narrative structure of the ego 18-20. Under DMT, this rhythm collapses 18. To record and quantify this collapse without MRI-induced noise, researchers utilize advanced electrophysiological pipelines:

EEG Preprocessing and Artifact Minimization

Recording EEG inside an active fMRI scanner introduces massive electromagnetic noise 21. Researchers resolve this via:

An MR-compatible cap referencing the FCz electrode with impedances kept strictly below 20 kΩ 21.

Gradient Artifact (GA) Removal: Employs an Average Artifact Template Subtraction (AAS) algorithm, subtracting a sliding-average template calculated across 21 TR windows 22.

Ballistocardiogram (BCG) Artifact Removal: Pulse artifacts from cardiac cycles are minimized by placing heartbeat markers on the ECG R-peak, generating a sliding-window subtraction template 22.

Irregularly Resampled AutoSpectral Analysis (IRASA)

A common confounding factor in spectral EEG is fractal (1/f) background noise 23. To isolate the true oscillatory power of the alpha band:

Researchers perform IRASA on the preprocessed signals to mathematically separate the 1/f power-law decay from the specific oscillatory components 23.

The isolated spectrum is divided into standard frequency bands, focusing on alpha (8–13 Hz) 23.

Average baseline activity (recorded eyes-closed outside the scanner pre-injection) is subtracted from both DMT and placebo conditions to eliminate order effects and overall differences in signal quality 23, 24.

Cluster-Based Permutation Testing: To locate regions of significant alpha collapse across the scalp, t-statistics are evaluated against a null distribution using 7,500 permutations 23. This reveals a profound, widespread collapse of posterior and parietal alpha power under DMT 18, 20.

EEG Signal Diversity (Lempel-Ziv Complexity)

To quantify the transition of the brain into a more entropic, complex state:

The raw EEG signal at each electrode is binarized using its mean value across 2-second epochs 24.

The Lempel-Ziv 1976 (LZ76) algorithm is applied to this binary stream to count the rate of unique subsequences, generating single-channel (LZs) and global channel-averaged (LZc) complexity metrics 24.

A dramatic increase in LZc under DMT correlates directly with subjective ratings of the "richness" of the conscious experience 7, 20.

Cortical Traveling Wave Analysis

To determine how the global flow of alpha rhythms is disrupted:

EEG signals are segmented into 1-second windows (sliding by 500 ms) 25.

A 2D time-electrode map is constructed by stacking five midline electrodes: Oz, POz, Pz, Cz, and FCz 25.

Performing a 2D Fast Fourier Transform (2D-FFT) on these maps isolates traveling waves 25. Under baseline conditions, alpha waves travel in a top-down, backward (BW) direction (from frontal to occipital regions) 20, 25. Under DMT, this backward traveling wave power collapses significantly, replaced by a surge in bottom-up, forward (FW) wave power 20.

3. Simultaneous Multimodal Fusion (EEG-fMRI Coupling)

To prove a direct neurobiological link between electrophysiological alpha collapse and fMRI DMN disintegration, researchers perform direct parallel trajectory correlation 26, 27:

Electrode Selection & Processing: Alpha power time series are extracted from the parietal electrodes showing the most pronounced statistical difference 27.

Hemodynamic Matching: To align the temporal scales of the two modalities, the EEG alpha time-series is band-pass filtered (0.01 to 0.08 Hz) and convolved with a canonical Hemodynamic Response Function (HRF) 27.

Gaussian Convolution: The resulting convolved EEG timecourse is smoothed with a 6-second Gaussian filter to match the sliding-window fMRI dynamic connectivity matrices 27.

Linear Mixed-Effects Modeling: Using data from participants surviving both fMRI and EEG head-motion/artifact thresholds, researchers run a linear mixed-effects model 27:

\\text{EEG alpha power}(t) \sim \text{fMRI GFC}(t) + (1 + \text{fMRI GFC} \mid \text{Subject})\

This LME framework directly exploits DMT-induced within-subject temporal trajectories 27.

The resulting multi-modal correlation model reveals that the localized collapse of parietal alpha power is mathematically and temporally coupled with a massive increase in GFC across high-level transmodal networks (the DMN and FP) 26. By demonstrating this tight coupling, researchers have confirmed that the disintegration of the default mode network and the desynchronization of the brain's alpha gate are dual facets of the same systemic transition to an entropic, anarchic state of consciousness 18, 26.

📊 Would you like us to run a custom statistical correlation in python using our simulated whole-brain Stuart-Landau network coordinates to evaluate how these GFC shifts mathematically alter the brain’s "small-world" topological efficiency?