Towards whole brain mapping of the haemodynamic response function, 2024, Mangini et al.

[SNT Gatchaman]

Towards whole brain mapping of the haemodynamic response function
Fabio Mangini; Marta Moraschi; Daniele Mascali; Maria Guidi; Michela Fratini; Silvia Mangia; Mauro DiNuzzo; Fabrizio Frezza; Federico Giove

Functional magnetic resonance imaging time-series are conventionally processed by linear modelling the evoked response as the convolution of the experimental conditions with a stereotyped haemodynamic response function (HRF). However, the neural signal in response to a stimulus can vary according to task, brain region, and subject-specific conditions. Moreover, HRF shape has been suggested to carry physiological information.

The BOLD signal across a range of sensorial and cognitive tasks was fitted using a sine series expansion, and modelled signals were deconvolved, thus giving rise to a task-specific deconvolved HRF (dHRF), which was characterized in terms of amplitude, latency, time-to-peak and full-width at half maximum for each task.

We found that the BOLD response shape changes not only across activated regions and tasks, but also across subjects despite the age homogeneity of the cohort. Largest variabilities were observed in mean amplitude and latency across tasks and regions, while time-to-peak and full width at half maximum were relatively more consistent. Additionally, the dHRF was found to deviate from canonicity in several brain regions.

Our results suggest that the choice of a standard, uniform HRF may be not optimal for all fMRI analyses and may lead to model misspecifications and statistical bias.

Link (Journal of Cerebral Blood Flow & Metabolism) [Open Access]

 

[SNT Gatchaman]

Potentially of relevance to fMRI studies, including FND studies and the TPJ findings in Deep phenotyping of post-infectious myalgic encephalomyelitis/chronic fatigue syndrome (2024, Nature Communications)

Introduction —

Functional Magnetic Resonance Imaging (fMRI) is an indirect measure of neuronal activity which leverages on the neurovascular coupling. The neuronal-driven vascular response gives rise to local changes of paramagnetic deoxyhaemoglobin concentration which can be detected with the Blood Oxygenation Level Dependent (BOLD) contrast. Neurovascular coupling is a complex and highly variable phenomenon which can be even disrupted in certain cases, for example in brain tumours, further complicating the study of neuronal activity.

Different approaches exist to infer brain activation from the BOLD response, including […] however, it is common practice to derive a statistical predictor by linear modelling the evoked BOLD response as the convolution of the known experimental conditions (onset, intensity, and duration of stimuli) with a stereotyped HRF [Haemodynamic Response Function]. The shape and features of the HRF, such as its magnitude, latency, and duration, have been reported to carry information on the underlying neuronal activity and depend on the interaction between complex neuronal, glial, and vascular events that lead from a stimulus to the measured BOLD response.

Discussion —

Here, we assessed the response variability across a wide range of sensorial and cognitive block-design tasks. These stimulations have been shown to elicit a positive BOLD response with a high level of significance across a large part of the cortex

Assessing the BOLD response can shed light on its role in physiological aging, cerebrovascular diseases, and dementia. Indeed, several studies have reported that multiple aspects of the HRF (e.g., amplitude, initial dip, post-stimulus undershoot) may be affected both by aging and neurodegenerative, psychiatric, and neurovascular diseases.

Typically, fMRI experiments are conducted on smaller sample sizes and investigate a restricted number of tasks and brain regions. […] In particular, aging is associated with vascular changes which can have an effect on the shape and timing of the HRF; this study focused on a homogeneous cohort of participants (i.e., healthy subjects aged 22–35 years) in order to reduce such variability. Age-related changes in the HRF have been observed, while no significant change in neurovascular coupling was found in association with age.

Latency was characterized by the highest variance across tasks, while TTP [Time-To-Peak] and FWHM [Full Width Half Maximum] were fairly constant across the tasks considered. This could be taken as an indication that TTP and FWHM are mostly driven by regional vascular features while latency, showing a stronger taskrelated variability, could hence be more sensitive to different extents of cognitive engagement.

The large spatial variability observed in latency, with occipital and temporal regions showing the fastest and slowest dynamics, respectively, could be due to different mechanisms. In particular, it could have a neuronal origin and, thus, indicate differences in brain activity, or it could be related to haemodynamic features, indicating primarily vascular differences, or a mixture of both. In the first case, the delay would represent a feature of interest for the characterization of brain activity following different tasks and for studies based on the temporal features of the BOLD response; in the second case, it would mainly represent a bias to correct for, if the target of the investigation is isolating the underlying neural activity.

From a practical point of view, the use of distinct HRFs corresponding to distinct tasks and areas would introduce a serious degree of complexity in fMRI analyses.

The spatial heterogeneity of the timing features of the dHRF is of particular relevance in functional connectivity applications. According to our findings, all timing parameters exhibit some degree of spatial variability. In particular, the mean latency of the BOLD response can have across-region variations in the order of several seconds (for example, up to 9 seconds between primary motor areas and temporal gyrus), that borders with the band of interest for functional connectivity. Functional connectivity based on correlation analysis may therefore miss the presence of existing connections between certain areas or underestimate their strength. In the case of time-lagged correlation analysis, strong timing variability could even lead to a misinterpretation of the directional relationship between brain regions. Given that the variability of the timing parameters both across tasks and subjects is relatively low, these biases might be systematically affecting the same regions, i.e., those characterized by larger differences.

Concludes —

We investigated HRF shape and timing parameters in the majority of the brain cortex and for a broad range of functional tasks using a deconvolution approach. BOLD response and dHRF shape and parametrization were found to vary both across subjects and across brain areas, with amplitude and latency showing a higher variability compared to FWHM and TTP.

Our finding suggests that the assumption of a single HRF can lead to biases in task-based and in connectivity fMRI studies, and that the use of flexible models is better suited for modelling the BOLD response when the exact shape of the response is expected to play a role.