Multimodal precision neuroimaging of the individual human brain at ultra-high field

2217 

Abstract Submission 

Donna Gift Cabalo¹, Ilana Leppert², Yezhou Wang¹, Risa Thevakumaran³, Shahin Tavakol⁴, Jessica Royer⁵, Jordan DeKraker⁶, Valeria Kebets⁷, Oualid Benkarim⁶, Bin Wan⁸, Youngeun Hwang⁷, Nicole Eichert⁹, Casey Paquola¹⁰, Sofie Valk⁸, Jonathan Smallwood¹¹, Christine Tardif², David Rudko², Raúl Rodriguez-Cruces⁶, Boris Bernhardt⁶

¹McGill University, Montreal, Quebec, ²McConnell Brain Imaging Centre, Montreal Neurological Institute and Hospital, Montreal, Quebec, ³McConnell Brain Imaging Centre, Montreal Neurological Institute and Hospital, Montreal, Montreal, Quebec, ⁴McGill University, Montreal Neurological Institute and Hospital, Montreal, Quebec, ⁵Montreal Neurological Institute and Hospital, Montreal, QC, ⁶Montreal Neurological Institute and Hospital, Montreal, Quebec, ⁷McGill University, Montreal Neurological Institute, Montreal, Quebec, ⁸Max Planck Institute for Human Cognitive and Brain Sciences, Leipzig, Germany, ⁹University of Oxford, Oxford, Oxfordshire, ¹⁰INM-7, Jülich, Jülich, ¹¹Department of Psychology, Queen’s University, Ontario, Canada

Donna Gift Cabalo  
McGill University
Montreal, Quebec

Ilana Leppert  
McConnell Brain Imaging Centre, Montreal Neurological Institute and Hospital
Montreal, Quebec

Yezhou Wang  
McGill University
Montreal, Quebec

Risa Thevakumaran  
McConnell Brain Imaging Centre, Montreal Neurological Institute and Hospital, Montreal
Montreal, Quebec

Shahin Tavakol  
McGill University, Montreal Neurological Institute and Hospital
Montreal, Quebec

Jessica Royer  
Montreal Neurological Institute and Hospital
Montreal, QC

Jordan DeKraker  
Montreal Neurological Institute and Hospital
Montreal, Quebec

Valeria Kebets  
McGill University, Montreal Neurological Institute
Montreal, Quebec

Oualid Benkarim  
Montreal Neurological Institute and Hospital
Montreal, Quebec

Bin Wan  
Max Planck Institute for Human Cognitive and Brain Sciences
Leipzig, Germany

Youngeun Hwang  
McGill University, Montreal Neurological Institute
Montreal, Quebec

Nicole Eichert  
University of Oxford
Oxford, Oxfordshire

Casey Paquola  
INM-7
Jülich, Jülich

Sofie Valk  
Max Planck Institute for Human Cognitive and Brain Sciences
Leipzig, Germany

Jonathan Smallwood  
Department of Psychology, Queen’s University
Ontario, Canada

Christine Tardif  
McConnell Brain Imaging Centre, Montreal Neurological Institute and Hospital
Montreal, Quebec

David Rudko  
McConnell Brain Imaging Centre, Montreal Neurological Institute and Hospital
Montreal, Quebec

Raúl Rodriguez-Cruces  
Montreal Neurological Institute and Hospital
Montreal, Quebec

Boris Bernhardt  
Montreal Neurological Institute and Hospital
Montreal, Quebec

Neuroimaging has advanced our understanding of the human brain by allowing non-invasive examination of brain structure and function. Nevertheless, human MRI research has predominantly centred around group-averaged data, which limits the specificity and clinical utility that MRI can offer[1]. Precision neuroimaging facilitates individualized mapping of brain structure and function through the use of repeated and prolonged scans[1]. A dense sampling of fMRI allows for detailed and reliable characterization of individual brain states and heteromodal networks[2]. Structural MRI's specificity can be augmented by using multiple quantitative MRI sequences, providing microstructural parameters characterizing inter-regional heterogeneity and inter-individual differences. Harnessing ultra-high field (UHF) neuroimaging at magnetic field strengths of 7 Tesla, can further enhance spatial and temporal resolution and is often imperative for precise mapping of highly susceptible and deep structures[3, 4]. Several initiatives have generated open-source UHF datasets; however, these focused either on functional[4] or structural sequences[3]. Here, we describe a multimodal precision neuroimaging dataset that capitalized on multiple sessions 7T MRI.

Our imaging protocol was implemented at the Montreal Neurological Institute and data were acquired on a 7T Terra Siemens scanner with the 8/32-channel transmit/receive Nova head coil. Ten healthy subjects (5M/5F, age=26.8±4.61, left/right handed=2/8) underwent three imaging sessions, each consisting of five distinct structural and five functional imaging protocols (Fig. 1A,C). Our UHF MRI data were processed with micapipe_v0.2.2[5], a surface-based processing software developed in my host lab, which introduced a standardized processing workflow for multiparametric UHF MRI acquisition. Structural scans included: (i) T1w and (ii) T1 relaxometry (T1) for examining intracortical microstructural organization, (iii) DWI for examining structural connectomes and fibre architectures, (iv) myelin-sensitive magnetization transfer, and, (v) iron-sensitive T2*-weighted multi-echo gradient echo. Each multi-echo fMRI scan lasted for 6 minutes and included, (i) rs-fMRI, (ii,iii,iv) multi-state task-based fMRI[6] with episodic encoding/retrieval and semantic tasks (Fig. 2B left). We also collected fMRI data as subjects watched (v) movies to study brain activity during naturalistic conditions[7]. Finally, we used the multidimensional experience sampling questionnaire (MDES), to study patterns of ongoing thought[7 ] (Fig. 2B right).

Alongside anonymized raw data, we will release fully processed[5] data for each modality, which provides surface-based neuroimaging features such as cortical thickness, quantitative MRI maps, and inter-regional structural and functional connectomes (Fig.1A,B). With this unprecedented dataset, we have generated upsampled data (i.e., from 0.5mm to 0.25mm isovoxels) from MRI acquisitions and averaged denoised data across sessions[8,9]. We have tested this method, particularly focusing on T1w images which substantially enhanced image contrast, and, with T1 maps which provided more granular intensity profiles than single-session data (Fig. 2A bottom left). The multi-session data also allows to assess test-retest reliability[10] of various MRI features. Exemplary analysis of DMN connectivity derived from rs-functional connectomes demonstrated high intra-subject reliability and inter-subject DMN uniformity, with strong identifiability[10] (d=1.95), indicating reliable and distinct DMN patterns from our UHF rs-fMRI data while preserving individual differences (Fig. 2A bottom right).

Our open-access precision UHF dataset promises to become a key resource for researchers aiming to advance our understanding of structure-function relationships in individual human brains and is instrumental in the development of novel image processing and analysis methodology.

Long-Term Memory (Episodic and Semantic)

Cortical Anatomy and Brain Mapping

Cortical Cyto- and Myeloarchitecture

Databasing and Data Sharing ¹

Multi-Modal Imaging ²

FUNCTIONAL MRI

HIGH FIELD MR

Open Data

STRUCTURAL MRI

Other - Ultra-high field 7T