You're not a brain in a vat.
You're a brain in a body.

	smoia
	@SteMoia
	s.moia.research@gmail.com

Tainan, 15.06.2025

Stefano Moia, 2025

The impact of dynamic and static
physiological sources on brain data.

Faculty of Psychology and Neuroscience, Maastricht University, Maastricht, The Netherlands; Open Science Special Interest Group (OHBM); physiopy (https://github.com/physiopy)

This is a new chapter

Take home #0

This is a take home message

Brains in vats?

No, brains in bodies

Impact of physiology on data variance

1. Bianciardi et al., 2009 (Magn. Reson. Imaging.); 2. Triantafyllou et al., 2005 (NeuroImage);

3. jorge et al., 2013 (Magn. Reson. Imaging.), Reynaud et al., 2017 (Magn. Reson. Med.)

Physiology-related variance varies:

By voxel size¹ ² and position¹
By field strength²
By sequence type and TR³

Impact of physiology on data variance

1. Krentz et al., 2023 (bioRxiv), Carlton et al., 2024 (bioRxiv), Moia et al., 2024 (bioRxiv);

2. Birn et al., 2009 (NeuroImage), image courtesy of Jingyuan Chen; 3. Lee et al., 2023 (HBM)

Physiology-related variance varies:

By individual & session¹
By task (task-locked)²
By Resting State Network³

Ventilation

Task (convolved)

RETROICOR variability

Neurovascular Coupling
and
Cerebrovascular Reactivity

Neurovascular Coupling

Neurovascular Coupling

Image courtesy of Dimo Ivanov

Changes in
Haemodynamics

Neurovascular
Coupling

Changes in Oxygen Metabolism

Changes in
BOLD signal

(Venous) Vasculature

Arrows indicate causal influence

Neurovascular Coupling

Image courtesy of Dimo Ivanov

Changes in
Haemodynamics

Neurovascular
Coupling

Changes in Oxygen Metabolism

Changes in
BOLD signal

(Venous) Vasculature

Arrows indicate causal influence

Cerebrovascular regulatory mechanisms, e.g. cerebrovascular reactivity

Cerebrovascular Reactivity and cognitive performance

Cerebrovascular Reactivity (CVR)

Cerebrovascular Reactivity (CVR) is the response of cerebral vessels to a vasoactive stimulus (e.g. CO2) to provide sufficient O2 to cerebral tissues¹

1. Liu et al., 2018 (Neuroimage); 2. Pinto et al., 2021 (Front. Physiol.), Moia et al., 2021 (Neuroimage)

CVR can be measured during BOLD fMRI experiments with Breath-holds (BH), that induce the subject into a state of hypercapnia²

This is not the golden standard method, but it is an affordable one.

Improving CVR quality

BH-induced CVR: Issues

Motion

CO2

BOLD

Lagged responses

Collinear motion

Subject compliance

1. Frederick et al. 2012 (NeuroImage); 2. Sousa et al. 2014 (Neuroimage)

CVR estimates optimisation

Different methods have been proposed to take into account the lagged CVR:

RIPTiDe: cross-correlation (X-corr)¹
Lagged GLM after orthogonalising signal of interest to noise²

Lag optimisation

Different methods have been proposed to take into account the lag of CVR:
cross-correlation (RIPTiDe)¹,
Lagged GLM (L-GLM)²,

bayesian estimation³, ...

1. Frederick et al. 2012 (NeuroImage); 2. Sousa et al. 2014 (NeuroImage)

3. https://github.com/physimals/quantiphyse-cvr; 4. Moia, Stickland, et al. 2020 (EMBC)

→

An important factor to take into account is to set up denoising and CVR estimation simultaneously⁴

[Moia, Stickland, et al. 2020 (EMBC)]

Methods: Analysis

Peaks detection in exhaled CO2 traces
Amplitude envelope (PETCO2) convolved with HRF
Find max correlation of PETCO2 and average GM
PETCO2 signal shifted between -9 and +9 sec
Shifted regressors interpolated at the TR
Compute GLM with each lagged regressor and nuisance regression (12 motion parameters and low frequency trends)
For each voxel, select the lagged model with highest explained variance (R²)

For comparison:

Computed non-optimised CVR map (measurement delay)
Computed GLM with each regressor, without motion parameters
Computed GLM with each regressor, after motion was regressed out

↓

optimised CVR map and lag map

Lag optimisation

1. Frederick et al. 2012 (Neuroimage), Sousa et al. 2014 (Neuroimage), Bayes et al., 2024 (biorXiv)
2. Moia, Stickland, et al. 2020 (EMBC), Moia et al. 2021 (Neuroimage)

→

Adopting a lagged GLM, maximising the R² of models including both signal of interest
and known noise, improves BH-based CVR estimation¹..

Subject compliance

1. Stickland et al., 2021 (NeuroImage); 2. Zvolanek et al., 2023 (Neuroimage)

Short respiratory challenges¹
Alternative estimates²

Easy additions!

Subject compliance

1. Stickland et al., 2021 (NeuroImage); 2. Zvolanek et al., 2023 (Neuroimage)

Short respiratory challenges¹
Alternative estimates²

Take home #1

Adding two Deep Breaths or Breath Holds before/after your task is an easy way
to estimate Cerebrovascular Reactivity

Solving the motion issue

"Advanced" modelling: ICA denoising

Independent Component Analysis (ICA) is commonly used to remove motion effects and other sources of noise from fMRI data

Motion + MB artefact

CSF pulsations

Griffanti et al. 2014 (NeuroImage), 2017 (NeuroImage), The tedana Community et al. 2021 (Zenodo)

Timeseries →

Spatial maps →

Power spectrum →

Advanced data acquisition: Multi-Echo (ME-fMRI)

Single Echo (SE): during fMRI acquisition we collect the signal once per TR at a certain TE to obtain one timeseries per voxel
Multi-Echo (ME): we collect the signal multiple times at different TEs to obtain n timeseries per voxel

BOLD [a.u.]

Text

~ BOLD

Multi-Echo improves contrast and tSNR

For each voxel and TR, we can Optimally Combine the echo volumes with a weighted sum based on their contribution to \( T_2^{\star} \)

In this way, spatial CNR and tSNR are maximised and the signal can be recovered in areas of drop-out

Posse et al. 1999 (Magn Reson Med), Poser et al., 2006 (Magn Reson Med)

Merging Multi-Echo and ICA denoising: tedana

DuPre et al., 2021 (JOSS)

ME-ICA

Assuming monoexponential decay, we can express the
signal percentage change as:

\[ S_{SPC} \approx \Delta\rho - TE \cdot \Delta R_2^{\star} + n \quad where \enspace R_2^{\star} = \frac{1}{T_2^{\star}} \]

This let us differentiate BOLD-related (\(\Delta R_2^{\star}\)) from non-BOLD related (\(\Delta\rho\)) changes

Kundu et al. 2012 (NeuroImage)

If we apply ICA, we can fit the timeseries of the components to either sub-models and automatically classify them

Multi-Echo & ICA based denoising

Reddy et al., 2024 (Imaging Neuroscience)

Increasing Motion

Single Echo →

Multi-Echo →

Multi-Echo & ICA →

Take home #2

Multi-Echo fMRI joint with denoising based on Independent Component Analysis can greatly reduce the impact of motion and other artefacts in your data

(Caveat: better at high fields than at ultra-high fields)

Which ICA denoising?

What is the best way to denoise BH data after ICA?

Being too aggressive might remove the signal of interest,
but being too conservative might keep too much noise in the model.

We tested the best approach by controlling the variance removed by the independent components

Motion removal

Moia et al. 2021 (NeuroImage)

Raw

Single Echo

Multi-Echo

Aggressive ME-ICA

Moderate ME-ICA

Conservative ME-ICA

Motion

Sequential vs Simultaneous denoising

Performing denoising in sequential steps, rather than in parallel, might reintroduce removed artefacts

Lindquist et al. 2019 (Hum. Brain Mapp.)

Results: lag optimisation

CNR of lag maps	SimMot	SeqMot	NoMot
GM-WM	0.52 ±0.21	0.46 ±0.26	0.49 ±0.25
GM-Putamen	0.47 ±0.22	0.44 ±0.21	0.44 ±0.21
GM-Cerebellum	0.82 ±0.15	0.69 ±0.17*	0.69 ±0.16*

Non optimising leads to underestimate the CVR, especially in subcortical areas.

Lag maps show anatomical consistency

→

Different lag responses, coherent with previous evidence (e.g. Putamen has earlier response than GM)

Area mostly affected by motion! →

Moia, Stickland, et al. 2020 (EMBC)

CVR amplitude and lag estimation and reliability

Moia et al. 2021 (NeuroImage)

CVR

Lag

CVR

Lag

Reliability (ICC)

Single Echo

Multi-Echo

Aggressive

Moderate

Conservative

Reliability [ICC (2,1)]

Moia et al. 2021 (NeuroImage)

Results: model comparison

X \sim model + (1|subject) + (1|session)

Moia et al. 2021 (NeuroImage)

CVR amplitude

CVR lag

CVR amplitude

CVR lag

Results: model comparison

X \sim model + (1|subject) + (1|session)

Moia et al. 2021 (NeuroImage)

CVR amplitude

CVR lag

CVR amplitude

CVR lag

Results: model comparison

X \sim model + (1|subject) + (1|session)

Moia et al. 2021 (NeuroImage)

CVR amplitude

CVR lag

CVR amplitude

CVR lag

Motion removal

Moia et al. 2021 (NeuroImage)

\( \cdot \) timepoint \( - \) session

Motion (FD)

Changes in BOLD (DVARS)

Raw

Single Echo

Multi-Echo

Aggressive ME-ICA

Moderate ME-ICA

Conservative ME-ICA

*Less slope is better

OC-MPR vs ME-CON

ME-CON should be better than OC-MPR in temporal-dependent application

Example of application: timeseries clustering (Self-Organising Maps, 20 clusters)

Submitted to ISMRM 2022

Take home #3

Pay attention to your ICA denoising:
an aggressive denoising might remove your signal of interest completely!

Alternative ICA based denoise (sequential)

What is the best way to denoise the data after ICA?

Regression based:

Aggressive approach: nuisance regression using only the rejected components.
Non aggressive (partial regression) approach: all the components are considered, but only the rejected components are regressed out of the data.
Orthogonalised approach: the rejected components are orthogonalised with respect to the other components.
4D-based approach (similar to M/EEG): reconstruct volumes on noise, then subtract it from the original data.

\( Y = \) \(A\) \(+\) \(R\) \(+ n \)

Multivariate:

Alternative ICA based denoise (sequential)

Effect of denoising approach is significant for slope (F(5,354)=177.6, p<0.001) and intercept (F(5,354)=225.7, p<0.001) of the linear regression model

DVARS vs FD

Group level, DVARS vs FD

Slope

Intercept

↓

Alternative ICA based denoise (sequential)

Average % BOLD and DVARS across all trials

↑

↓

Aggressive denoise removes signal of interest

OC and E-02 denoise affects the signal of interest more than ICA denoise

↑

Alternative ICA based denoise (sequential)

Lag reliability

CVR reliability

Impact of CVR
on rest and behaviour

Cerebrovascular Reactivity (CVR) and Neurovascular Coupling

CVR reflects the capacity for Cerebral Blood Flow and Volume to increase.

A similar, if not the same, process is driven by Neurovascular Coupling.

Does CVR amplitude predict Neurovascular Coupling?

Is CVR a predictor of natural fluctuations and behavioural responses?
Does it explain part of the individual variability?

An alternative estimation of CVR

Previous literature links CVR with Resting State Fluctuations Amplitudes, due to cross-sectional (inter-subject) and spatial correlation¹

RSFA = \sigma_{t_x}

CVR \backsim RSFA

1. Golestani et al., 2016 (Neuroimage), Chen et al., 2023 (bioRxiv), De Vis et al. 2014 (PloS ONE), Kannurpatti et al., 2014 (PloS ONE)

Alternative CVR estimation

Resting State Fluctuations (RSFA, ALFF) agree with CVR in cross-section studies¹

1. Golestani et al., 2016 (Neuroimage), Chen et al., 2023 (bioRxiv),
De Vis et al. 2014 (PloS ONE), Kannurpatti et al., 2014 (PloS ONE)

corr(CVR,RSFA)

Resting State Fluctuations and physiology

1. [adapted from] Power et al. 2017 (Neuroimage)

This susceptibility can be treated either as noise or as meaningful information (e.g. physiological imaging or functional MRI calibration).

Resting State Fluctuations (RSF) are susceptible to physiological signals¹.

Resting State Fluctuations and behavioural responses

CVR \backsim RSFA

RSFA → beh.resp.

CVR → beh.resp. \; {?}

1. Zou et al. 2013 (Hum. Brain Mapp.), Mennes et al. 2011 (NeuroImage), Kannurpatti et al. 2014 (PLoS ONE)

Link between Resting State Fluctuations Amplitudes and behavioural responses¹

* ALFF is qualitatively the same as RSFA

EuskalIBUR

ME-MB fMRI: TR = 1.5 s, TEs = 10.6/28.69/46.78/64.87/82.96 ms, MB = 4, GRAPPA = 2, voxel size = 2.4x2.4x3 mm³

10 neurotypical subjects
(5F, age 25-40y)

10 sessions, one week apart, same time of the day

T2w, MP2RAGE, 4 RS,
3 tasks, BH paradigm

CO2 and O2
sampling

↑

Moia et al. (2020), OpenNeuro. [Dataset] doi: 10.18112/openneuro.ds003192.v1.0.1

EuskalIBUR

ME-MB fMRI: TR = 1.5 s, TEs = 10.6/28.69/46.78/64.87/82.96 ms, MB = 4, GRAPPA = 2, voxel size = 2.4x2.4x3 mm³

10 neurotypical subjects
(5F, age 25-40y)

10 sessions, one week apart, same time of the day

T2w, MP2RAGE, 4 RS,
3 tasks, BH paradigm

CO2 and O2
sampling

↑

Moia et al. (2020), OpenNeuro. [Dataset] doi: 10.18112/openneuro.ds003192.v1.0.1

Impact of CVR on rest and behavioural responses

CVR \backsim RSFA

RSFA → beh.resp.

CVR → beh.resp. \; {?}

Data

Session-wise GLMs to retrieve betas and contrasts in tasks
Linear Mixed Effects Models to assess relationships between CVR, RSF, and behavioural responses

Motor

Simon

CVR, Resting State Fluctuations, and behaviour

Results thresholded at \(p<0.001\) uncorrected

CVR-
RSFA

Motor

Simon
Congr.

Simon
Incon.

Results: CVR and RSF

Results thresholded at \(p<0.001\) uncorrected

Results: CVR, RSF, and tasks (motor)

Results thresholded at \(p<0.001\) uncorrected

Results: CVR, RSF, and tasks (Simon, congruent)

Results thresholded at \(p<0.001\) uncorrected

Results: CVR, RSF, and tasks (Simon, incongruent)

Results thresholded at \(p<0.001\) uncorrected

Physiology and RSF

We used 3dLMEr¹ to set up the following LME models (R syntax):

RSF \sim sex*(MAP+PP+pulse)+(1|ses)+((MAP+PP+pulse)|subj)

Effect of Sex

Only sex had a significant effect on RSF

Effect on CVR

CVR and Resting States Fluctuations

Resting State Fluctuations (RSF) agree with CVR in cross-section studies¹
RSF do not agree strongly with CVR in a dense mapping study²
RSF might be more related to Neurovascular Coupling than to CVR³

1. Golestani et al., 2016 (Neuroimage), Chen et al., 2023 (bioRxiv); 2. Moia et al., 2024 (bioRxiv); 3. Bailes et al., 2023 (eLife)

Alternative CVR estimation?

Alternative CVR estimation

In the same cohort, we selected the most idiosyncratic connections at rest (i.e. \(ICC[1,1] \geqslant P_{99}ICC\)) and we used them to explain the incongruent effect Response Time (RT) in the Simon task
We observe the same subject-specific relationship between FC and RT

Idiosyncratic functional connectivity and behaviour

Take home #4

Inter-individual variability reveals variable relationships between CVR and
BOLD fluctuations at rest (Neurovascular Coupling)
as well as between the latter and behaviour.

Vessels
and
Multimodal Imaging

Physiological and vascular networks

BOLD-based fMRI, in conjunction with physiological signals, reveals the existence of physiological and vascular brain networks¹.

1. Bright et al. 2020 (NeuroImage), Chen et al., 2020 (NeuroImage)

Subject-wise analyses on Breath-Holds
Data driven algorithms (Self-Organising Maps) to cluster voxels in 15 partitions

Physiological and vascular networks in EuskalIBUR

SOM result, single subject

[adapted from] Moia et al. 2021 (Neuroimage)

Respiratory
challenges

Physiological networks

Impact of vessels on structural and functional connectivity

INDIVIDUAL (NSD)

GROUP (HCP)

1. Allen et al., 2022 (Nat. Neurosci.) 2. Glasser et al., 2013 (NeuroImage); 3. Huck et al., 2019 (Brain Struct. Func.)

Natural Scenes Dataset (NSD)¹ (T1w, T2w, ToF)

Human Connectome Project (HCP)² and the VENAT atlas³

Impact of vessels on structural and functional connectivity

GROUP & INDIVIDUAL

INDIVIDUAL

Impact of vessels on structural and functional connectivity

1. Glasser et al., 2016 (Nat.); 2. Moia er al., 2023 (Zenodo)

GROUP (HCP)

INDIVIDUAL (NSD)

Functional Connectivity

Structural
Connectivity

Impact of vessels on structural and functional connectivity

1. Griffa et al., 2022 (NeuroImage), Moia er al., 2023 (Zenodo), Preti & Van De Ville, 2019 (Nat. Commun.)

GROUP (HCP)

INDIVIDUAL (NSD)

Functional Connectivity

Structurally-coupled FC¹

Structurally-decoupled FC¹

Impact of vessels on structural and functional connectivity

GROUP

INDIVIDUAL

Impact of vessels on structural and functional connectivity

Weak correlation between vascular properties and structural and functional connectivity at the group level
Negligible correlation at the individual level, and weak correlation with vessel distance
Analysis limited to macro scale vessels and non-pial veins.

"Static" physiology: vessels

1. Gulban et al., in prep., Gulban et al., 2025 (bioRxiv), image courtesy of Faruk Gulban; 2. Zhong et al., 2024 (Imaging Neurosci.)

Meso-scale vessels introduce partial volumes effects
in the grey matter¹
Macro-scale vessels affect connectivity
in the grey matter²

Take home #5

Macro- (and Meso-) scale vessels affect local (but not global) fluctuations in BOLD responses.

Working with Physiology

Physiopy in a nutshell

A set of easily adoptable toolboxes

Community of users, developers, and researchers interested in physiology

Clear and approachable documentation

Community practices based on consensus

physiopy

Raw data

BIDSification

phys. data preprocessing
(peak detect.)

physiopy's documentation
&

phys. denoising

phys. imaging

BIDS Extension Proposal

Physiopy's Community Practices

QA/QC

	github.com/physiopy
	physiopy.github.io physiopy-community-guidelines.rtfd.io
	physiopy.community@gmail.com s.moia.research@gmail.com

Toolboxes guarantee reproducibility even in presence of manual steps
Community guidelines guide new (and old) users in improving the quality and usage of physiological signals in neuroimaging

physiopy-community-guidelines.rtfd.io

That's all folks!

Part of this research is supported by the European Union’s Horizon Europe research and innovation programme under the Marie Skłodowska-Curie grant agreement No 101109770

Part of this research supported by the Eunice Kennedy Shriver National Institute of Child Health and Human Development of the National Institutes of Health under award number K12HD073945, the European Union’s Horizon 2020 research and innovation program (Marie Skłodowska-Curie grant agreement No. 713673), a fellowship from La Caixa Foundation (ID 100010434, fellowship code LCF/BQ/IN17/11620063), the Spanish Ministry of Economy and Competitiveness (Ramon y Cajal Fellowship, RYC-2017- 21845), the Spanish State Research Agency (BCBL “Severo Ochoa” excellence accreditation, SEV- 2015-490), the Basque Government (BERC 2018-2021 and PIBA_2019_104), the Spanish Ministry of Science, Innovation and Universities (MICINN; FJCI-2017-31814)

Any question [/opinions/objections/...]?

	smoia
	@SteMoia
	s.moia.research@gmail.com

Stefano Moia, 2025

Find the presentation at:

slides.com/smoia/you-re-not-a-brain-in-a-vat-ncku/scroll

You're not a brain in a vat [NCKU]

By Stefano Moia

You're not a brain in a vat [NCKU]

CC-BY 4.0 Stefano Moia, 2025. Images are property of the original authors and should be shared following their respective licences. This presentation is otherwise licensed under CC BY 4.0. To view a copy of this license, visit https://creativecommons.org/licenses/by/4.0/

You're not a brain in a vat. You're a brain in a body.

The impact of dynamic and static physiological sources on brain data.

This is a new chapter

This is a new chapter

Take home #0

This is a take home message

Brains in vats?

No, brains in bodies

Impact of physiology on data variance

Impact of physiology on data variance

RETROICOR variability

Neurovascular Coupling and Cerebrovascular Reactivity

Neurovascular Coupling

Neurovascular Coupling

Neurovascular Coupling

Cerebrovascular regulatory mechanisms, e.g. cerebrovascular reactivity

Cerebrovascular Reactivity and cognitive performance

Cerebrovascular Reactivity (CVR)

Improving CVR quality

BH-induced CVR: Issues

Lagged responses

Collinear motion

Subject compliance

CVR estimates optimisation

Lag optimisation

Methods: Analysis

Lag optimisation

Subject compliance

Subject compliance

Take home #1

Adding two Deep Breaths or Breath Holds before/after your task is an easy way to estimate Cerebrovascular Reactivity

Solving the motion issue

"Advanced" modelling: ICA denoising

Advanced data acquisition: Multi-Echo (ME-fMRI)

Multi-Echo improves contrast and tSNR

Merging Multi-Echo and ICA denoising: tedana

ME-ICA

Multi-Echo & ICA based denoising

Take home #2

Multi-Echo fMRI joint with denoising based on Independent Component Analysis can greatly reduce the impact of motion and other artefacts in your data

(Caveat: better at high fields than at ultra-high fields)

Which ICA denoising?

Motion removal

Sequential vs Simultaneous denoising

Results: lag optimisation

CVR amplitude and lag estimation and reliability

CVR

Lag

CVR

Lag

Reliability (ICC)

Reliability [ICC (2,1)]

Results: model comparison

Results: model comparison

Results: model comparison

Motion removal

OC-MPR vs ME-CON

Take home #3

Pay attention to your ICA denoising: an aggressive denoising might remove your signal of interest completely!

Alternative ICA based denoise (sequential)

Alternative ICA based denoise (sequential)

Alternative ICA based denoise (sequential)

Alternative ICA based denoise (sequential)

Impact of CVR on rest and behaviour

Cerebrovascular Reactivity (CVR) and Neurovascular Coupling

An alternative estimation of CVR

Alternative CVR estimation

Resting State Fluctuations and physiology

Resting State Fluctuations and behavioural responses

EuskalIBUR

EuskalIBUR

Impact of CVR on rest and behavioural responses

Data

Motor

Simon

CVR, Resting State Fluctuations, and behaviour

CVR- RSFA

Motor

Simon Congr.

Simon Incon.

You're not a brain in a vat.
You're a brain in a body.

The impact of dynamic and static
physiological sources on brain data.

Neurovascular Coupling
and
Cerebrovascular Reactivity

Adding two Deep Breaths or Breath Holds before/after your task is an easy way
to estimate Cerebrovascular Reactivity

Pay attention to your ICA denoising:
an aggressive denoising might remove your signal of interest completely!

Impact of CVR
on rest and behaviour

CVR-
RSFA

Simon
Congr.

Simon
Incon.

Inter-individual variability reveals variable relationships between CVR and
BOLD fluctuations at rest (Neurovascular Coupling)
as well as between the latter and behaviour.

Vessels
and
Multimodal Imaging

Structural
Connectivity