Development of MRI Technologies for Quantitative Measurements of Metabolic Dynamics, Oxygen Consumption, and Vascular Reactivity: Applications to the Study of Cerebral Energetics

We have fine-tuned the method for measuring cerebrovascular reactivity (CVR) and are currently acquiring data from Alzheimer’s patients. Over the next three years, we’ll first verify whether CVR provides useful insights into neurodegenerative processes. In a second phase, we will further develop the CVR measurement method to derive quantitative measures of cerebral metabolic rate of oxygen (CMRO2).

CMRO2 measurements will be combined with spectroscopy measurements to characterize the energetics of perception. We recently showed that visual perception induces a decoupling between the functional and metabolic responses; we aim to verify if this decoupling is linked to a different regulation of aerobic metabolism. This would have important consequences for interpreting functional data and understanding pathologies or conditions that impact perception (e.g., hallucinatory states). We also plan to study the association between metabolic modulations and the dynamics of the BOLD signal, which is a potential biomarker for numerous diseases.

This section was funded by the Lazio Region (FISASMEM project) and is partially funded by a PRIN-PNRR project (MUR PRIN 2022 P202294JHK “RECENTRE”).

Development of Heteronuclear Imaging with 23Na for Investigating Early Functional Alterations in Alzheimer’s Disease (AD)

We have recently developed MRI techniques based on 23Na imaging to identify potential biomarkers in AD and explore the pathophysiological processes behind microstructural tissue damage and cognitive decline. Sodium homeostasis is associated with neuroinflammation, with potential sensitivity to vascular and metabolic alterations. The methodological development has been completed, and data acquisition is underway, set to be finalized during the year. Using this data, we will test the association between sodium metrics, inflammatory/neurodegenerative processes, and cognitive dysfunction in AD to characterize the usefulness of 23Na metrics as biomarkers.

This section is currently funded by the PNRR through a synergistic project at the Santa Lucia Foundation (PNRR MAD-2022-12376889).

Characterization of Cerebral Network Dynamics and Identification of Non-Neuronal Components

Connectome analysis techniques are based on assessing data covariance structure and are highly sensitive to coherent spurious signals, including “physiological noise” (i.e., variations induced by physiological rhythms like breathing, movement, or heartbeat). In the near future, we will finalize our study on noise mitigation in high-resolution functional connectivity. Functional connectivity between cortical layers has the potential to evaluate the directionality of connections, distinguish input and output activity, and more directly investigate the function of microcircuits. However, the quality of high-resolution fMRI data is strongly degraded by noise. We have developed a denoising approach that combines spatial realignment of fMRI volumes, thermal denoising, and methods for mitigating physiological noise. Our results show a different effect of various noise components in cortical layers and a corresponding modulation of spurious intracortical connectivity.

Starting in 2025, this section will be substantially expanded, also converging with the “Development of MRI Technologies” section. We aim to provide a complete characterization of the functional aspects of pial vein dynamics in terms of their salient features, transients, and frequency composition. The goal is to push the limits of functional MRI at 3 Tesla up to 1 mm3, improve thermal and physiological denoising techniques, and provide new biomarkers for cerebrovascular health. The behavior of pial veins will be modulated in the experimental phase by administering simple functional tasks of different durations and types, while the vessels’ baseline state will be modulated through mild hypercapnia and hyperoxia. We will extract physiological and metabolic indicators as a function of the cortical layer from fMRI data and separate the vascular component. We will combine consolidated imaging methods (gradient-echo BOLD) with less conventional contrasts (VASO, or Vascular Space Occupancy for blood volume mapping, and pCASL for blood flow) and different experimental conditions (hyperoxia, hypercapnia, visual tasks, rest state). The respiratory challenges will allow for the calibration of the MRI signal and thus enable us to estimate CMRO2 variations in different cortical layers. However, since the fMRI signal at a certain depth does not necessarily reflect the activity of a single layer, it is essential to consider the cortical circuitry and the task-specific metabolic demand. To this end, we will use a consolidated cortical model, which incorporates non-linear synaptic plasticity, to predict the expected profile of CMRO2 variations during visual tasks, comparing it with the acquired fMRI data.

While the small size of the voxels will cause high thermal noise, which does not depend on cortical depth, physiological noise has a significant contribution in the upper layers. We will continue to optimize denoising to maximize noise uniformity within the gray matter and reduce spurious connectivity, which our results suggest primarily affects the upper layers. In the continuation of this study, we will verify if these denoising techniques allow for high-resolution fMRI at a clinical field strength (3 T) with acquisition times compatible with patient use. The duration of this project section is estimated at three years.

This section is currently funded by the PNRR through project M4 C2, “MNESYS SINVASC.”

Development of Spinal Cord Imaging

Although clinical MRI of the spinal cord is commonly used, it has so far proven unable to provide reliable quantitative metrics for a comprehensive characterization of tissue damage. In this context, we have optimized an experimental protocol and data analysis pipeline for the spinal cord to study patients with traumatic and inflammatory spinal cord injuries, both at the cervical and dorsal levels. We are currently applying the optimized protocol to healthy subjects and patients with cervical or dorsal lesions. To characterize the extent of the damage and predict clinical outcomes, lesion segmentation is a necessary step, which we are now refining. Lesion segmentation will allow for a precise localization of the damage, enabling the local calculation of advanced morphometric measures.

This section is currently funded by the PNRR with a synergistic project at the Santa Lucia Foundation (MCNT2-2023-12378303).

Characterization of the White Matter Connection Architecture and Related Microstructural Properties to Study Progressive Changes in the Onset of Alzheimer’s Disease

The white matter tracts in the human brain form the fundamental substrate for the integration of different gray matter areas and create a true infrastructure that underpins brain function. Damage to these tracts, whether localized in circumscribed areas (lesions) or diffused throughout entire anatomical structures, is central to the pathophysiology of numerous neurological diseases. Diffusion MRI makes it possible not only to reconstruct the short- and long-range connections that form the brain’s network but also to characterize the microscopic properties of the related tissues through compartmental models that rely on an extensive use of diffusion gradients.

Over the next three years, we will characterize the macro- and microstructure of white matter in different patient cohorts that cover a large part of the cognitive deficit spectrum: patients with subjective cognitive decline, those with mild cognitive impairment, and finally those affected by AD. This study will explore, in parallel, the onset and characterization of lesions and the localized differences in white matter structures associated with cognitive functions of interest.