Measuring the awake infant brain
Most of what is known about the infant brain comes from scanning babies asleep. That tells you about structure and intrinsic connectivity, but not about what the brain is doing while an infant looks at the world. Over the past seven years we have built the methods to scan infants awake, at scale. In the Foundations of Cognition project we scanned 134 infants at 2 and 9 months while they watched naturalistic videos and viewed object images. Ninety-seven percent of two-month-olds and 63% of nine-month-olds gave usable data, with a median of 13–19 minutes of scanning per infant, and both ages showed reliable stimulus-driven responses across the developing brain. What we learned about paradigm design, scheduling, soothing, attention monitoring and data quality is written up in Dineen et al. (2026), Frontiers in Neuroscience, which we hope will be useful to other labs starting out. And it wouldn't have been possible without TCIN's Siemens Prisma MRI.
Most of what is known about the infant brain comes from scanning babies asleep. That tells you about structure and intrinsic connectivity, but not about what the brain is doing while an infant looks at the world. Over the past seven years we have built the methods to scan infants awake, at scale. In the Foundations of Cognition project we scanned 134 infants at 2 and 9 months while they watched naturalistic videos and viewed object images. Ninety-seven percent of two-month-olds and 63% of nine-month-olds gave usable data, with a median of 13–19 minutes of scanning per infant, and both ages showed reliable stimulus-driven responses across the developing brain. What we learned about paradigm design, scheduling, soothing, attention monitoring and data quality is written up in Dineen et al. (2026), Frontiers in Neuroscience, which we hope will be useful to other labs starting out. And it wouldn't have been possible without TCIN's Siemens Prisma MRI.
|
Motion: the central problem
Awake infants move. In the FreezeMotion project, funded by a Research Ireland Frontiers for the Future award, we are developing deep-learning tools that correct the distortions head motion introduces into fMRI, including self-supervised slice-to-volume reconstruction with Chen Qin at Imperial College London (Wang et al., 2026, ). Current work on the project: Param Ahir, Muhammad Faizan |
|
Fast brain activity: OPM-MEG
fMRI has good spatial resolution but is slow. To see the dynamics of infant brain activity we are establishing awake infant OPM-MEG on the Institute's CERCA system, a new capability for which we are currently recruiting a postdoctoral fellow.
fMRI has good spatial resolution but is slow. To see the dynamics of infant brain activity we are establishing awake infant OPM-MEG on the Institute's CERCA system, a new capability for which we are currently recruiting a postdoctoral fellow.
Why this matters clinically
Better tools for imaging infants are also better tools for detecting when development has been disrupted. Much of the methods work here is motivated by the neonatal brain injury programme described on the Brain injury and the clinic page.
Code
Our analysis stack is entirely Python (NiPype, PyTorch, BIDS). Public code includes foundcog_analysis, the preprocessing and GLM pipelines for the FOUNDCOG awake infant data, and smartontheinside, the analysis code for Caldinelli & Cusack on structural connectivity in neonates.
Better tools for imaging infants are also better tools for detecting when development has been disrupted. Much of the methods work here is motivated by the neonatal brain injury programme described on the Brain injury and the clinic page.
Code
Our analysis stack is entirely Python (NiPype, PyTorch, BIDS). Public code includes foundcog_analysis, the preprocessing and GLM pipelines for the FOUNDCOG awake infant data, and smartontheinside, the analysis code for Caldinelli & Cusack on structural connectivity in neonates.
