The Athinoula A. Martinos Center for Biomedical Imaging, usually referred to as just the "Martinos Center," is a major hub of biomedical imaging technology development and translational research. Bruce Rosen is the Director of the Center and Monica Langone is the Administrative Director.The core technologies being developed and used at the Center are magnetic resonance imaging (MRI) and in vivo magnetic resonance spectroscopy (MRS), magnetoencephalography (MEG) and electroencephalography (EEG), optical imaging techniques (microscopy and near-infrared spectroscopy), positron emission tomography (PET), molecular imaging, and transcranial magnetic stimulation. A particular area of innovation at the Center is Multimodal Functional Neuroimaging, which involves the integration of imaging technologies for neuroscience applications. Major areas of research at the Center include: psychiatric, neurologic and neurovascular disorders; basic and cognitive neuroscience; cardiovascular disease; cancer; and more. Scientific investigation and technology development is funded through government, industry and other research grants.The center is located in the Massachusetts General Hospital (MGH) East Campus in the Charlestown Navy Yard, 149 13th St. Charlestown, MA 02129. Separately, Massachusetts Institute of Technology (MIT) is home to its own Martinos Imaging Center.The Martinos Center is home to approximately 120 faculty members and more than 100 postdoctoral research fellows and graduate students, and is a resource to hundreds of researchers and students throughout Boston, the United States and the world. The research faculty are basic scientists and clinicians interested in a broad range of biologically and medically important questions. They work in conjunction with the physical scientists and engineers to develop new imaging technologies and research applications, and to bring these developments to the sphere of medical care. Some of the prominent faculty at the Center include Bruce Rosen, Lawrence Wald, David Boas, Jacob Hooker, Julie C. Price, Peter Caravan, Anna Moore, Umar Mahmood, Randy Buckner, Matthew S. Rosen, Maria Angela Franceschini and Bruce Fischl.The Center includes investigators and their laboratories based at the MGH research campus in Charlestown, as well as numerous other researchers from various departments within MGH, and other local, national and international institutions. Most Martinos Center-based faculty members have primary appointments in Radiology at MGH, some with secondary appointments at MIT. Several of the investigators from other MGH departments and other institutions work at the Center, while even more conduct long- and short-term imaging studies at the Center and maintain their base elsewhere.The center is a member or collaborator with NCRR (and BIRN), NIDA, NIBIB, National Cancer Institute, NINDS, NCCAM, ONDCP, and The MIND Institute. The center also has a strategic corporate partnerships with Siemens Medical Solutions, Pfizer Inc., and Canon Inc. It is also a Harvard Catalyst site, and incorporates research projects from Boston University, McLean Hospital, and other Boston institutions.At the MGH Navy Yard site, there are eight large bore and five small bore MRI scanning bays used primarily for research, including the high-gradient field Human Connectome Project scanner, a 7 Tesla magnet for human radiography, and a combined PET-MRI. The Martinos Center also served as the site for the development of magnetoencephalography (MEG), and software development for analysis of MEG data is ongoing at the facility. New MRI and MRS sequences are developed in conjunction with Martinos, Harvard, and MIT faculty. In addition, the Center serves as a development site for new Siemens equipment, such as 32, 64, and 128 channel MRI coils which were designed and prototyped there..
We present FOMO260K, a large-scale, heterogeneous dataset of 260,927 brain Magnetic Resonance Imaging (MRI) scans from 77,589 MRI sessions and 55,378 subjects, aggregated from 910 publicly available sources. The dataset includes both clinical- and research-grade images, multiple MRI sequences, and a wide range of anatomical and pathological variability, including scans with large brain anomalies. Minimal preprocessing was applied to preserve the original image characteristics while reducing entry barriers for new users. Companion code for self-supervised pretraining and finetuning is provided, along with pretrained models. FOMO260K is intended to support the development and benchmarking of self-supervised learning methods in medical imaging at scale.
MRI is an indispensable clinical tool, offering a rich variety of tissue contrasts to support broad diagnostic and research applications. Protocols can incorporate multiple structural, functional, diffusion, spectroscopic, or relaxometry sequences to provide complementary information for differential diagnosis, and to capture multidimensional insights into tissue structure and composition. However, these capabilities come at the cost of prolonged scan times, which reduce patient throughput, increase susceptibility to motion artifacts, and may require trade-offs in image quality or diagnostic scope. Over the last two decades, advances in image reconstruction algorithms-alongside improvements in hardware and pulse sequence design-have made it possible to accelerate acquisitions while preserving diagnostic quality. Central to this progress is the ability to incorporate prior information to regularize the solutions to the reconstruction problem. In this tutorial, we overview the basics of MRI reconstruction and highlight state-of-the-art approaches, beginning with classical methods that rely on explicit hand-crafted priors, and then turning to deep learning methods that leverage a combination of learned and crafted priors to further push the performance envelope. We also explore the translational aspects and eventual clinical implications of these methods. We conclude by discussing future directions to address remaining challenges in MRI reconstruction. The tutorial is accompanied by a Python toolbox (https://github.com/tutorial-MRI-recon/tutorial) to demonstrate select methods discussed in the article.
Brain ageing involves microstructural changes that vary across tissue types and even within regions of those tissues, leading to functional and cognitive alterations. Quantitative MRI (qMRI) offers sensitivity to tissue properties, enabling the identification of differential ageing patterns and distinguishing physiological ageing from pathological changes. In this study, we analysed qMRI data from 293 healthy adults (median age: 52; interquartile range: 36-66; age range: 18-79 years). We applied a multiparametric qMRI approach, including longitudinal relaxation rate (R 1), apparent transverse relaxation rate (R 2*) and Quantitative Susceptibility Mapping, to model normal ageing effects on qMRI metrics across regions using second-order polynomial regression, adjusting for sex, education and cognition. Peak ages in turning points derived from quadratic fits were extracted to capture region-specific age-related differences across cortical grey matter, superficial white matter (sWM) and white matter (WM) bundles. According to the results, R 1 showed the most robust age modelling, whereas R 2* and susceptibility presented greater regional variability. Peak ages varied substantially across regions, reflecting the heterogeneity of age-related microstructural differences. Based on quadratic fits, we identified a spatial gradient in qMRI ageing patterns, with earlier peak ages in WM bundles, followed by sWM and culminating in cortical GM. This gradient followed a posterior-to-anterior pattern in the cortex and an inferior-to-superior pattern in WM bundles, consistently observed across all three qMRI metrics. Our study presents exploratory mapping of region- and tissue-specific ageing patterns across brain grey and WM using multiparametric qMRI, offering insights to support future normative healthy ageing research.
Functional MRI-based graph theory has provided profound insights into the brain's functional organization, yet the neuroenergetic meaning of widely used graph-theoretical metrics remains poorly understood. Although resting-state research suggests a positive coupling between network topology and glucose metabolism, it remains unclear whether this relationship reflects a general principle of brain organization or a state-specific phenomenon. Here, we test the neuroenergetic interpretability of nodal graph-theoretical metrics by linking complex network topology to cerebral glucose consumption across diverse brain states. Leveraging simultaneous functional PET-MRI, we directly compare state-dependent fluctuations in glucose consumption and network topology during sensory, cognitive, and arousal conditions. We further assess metabolic-topological couplings in disease through a meta-analysis of resting-state FDG-PET and fMRI studies involving Alzheimer's disease, Parkinson's disease, major depressive disorder, and schizophrenia. Our results show that nodal graph-theoretical metrics exhibit state- and network-dependent metabolic associations, with coupling patterns diverging across experimental and disease contexts. Notably, frontoparietal and attentional networks show more conserved metabolic-topological coupling than other large-scale networks across states. These findings underscore a dynamic, complex interplay between metabolic demand and complex network organization, highlighting the need for a nuanced interpretation of the energetic underpinnings of nodal graph-theoretical metrics in health and disease.
