Biology subjects

Baljer, L.

Publications and source records attributed to Baljer, L..

4 recordsLinked to original sources

bioRxiv2024

Deep learning super-resolution of paediatric ultra-low-field MRI without paired high-field scans

Brain magnetic resonance imaging (MRI) is essential for diagnosis and neurodevelopmental research, but the high cost and infrastructure demands of high-field MRI limit its use to high-income settings. Ultra-low-field MRI scanners offer a more affordable and energy-efficient alternative, but their reduced resolution and signal-to-noise ratio restrict research and clinical utility, prompting the need for super-resolution techniques. Current super-resolution methods rely on either three anisotropic ultra-low-field scans acquired at different orientations (axial, coronal, sagittal) to reconstruct a higher-resolution image using multi-resolution registration (MRR) or the training of deep learning models using paired ultra-low- and high-field scans. Since acquiring three high-quality ultra-low-field scans is not always feasible, and paired high-field data may not be available, this study explores the efficacy of using a deep learning model to generate scans of MRR quality from a single ultra-low-field input scan. Results demonstrated significant enhancement in the quality of output scans, including improved image quality metrics, stronger tissue volume correlations, and greater Dice overlap of tissue segmentations. Generating higher-resolution brain scans from single ultra-low-field scans, without paired high-field data, reduces scanning time and further widens MRI accessibility in low- and middle-income countries. This approach also facilitates site-specific model training, which an exploratory external validation suggests may be necessary to address potential domain shifts across scanning sites.

neuroscience↗

bioRxiv2024

Computational neurodevelopment: infant decision-making in changing environments

In recognition of the fact that most psychiatric conditions have neurodevelopmental origins, there is an increasing interest in applying the methodological and conceptual approaches from computational psychiatry to developmental cohorts. However, the challenge of acquiring and modelling behavioural responses in very young infants has thus far proven difficult to overcome. To address this we developed a novel gaze-contingent, cued-reversal paradigm that allowed 6-10 month old infants to make overt behavioural responses to assess learning of expectations and updating of behaviour in response to change. We then fit computational models to infant behaviour and, for the first time, were able to validate the winning model to the same standards as would be expected of adults (e.g. good parameter recoverability, model identifiability and simulated behavioural responses). Similar to prior findings in adults, model-based prediction error measures correlated with post-switch increases in pupil size; consistent with noradrenalines hypothesised role in learning about change. Data-driven clustering based on model parameters revealed two infant behavioural subtypes hidden within the data; one with a perseverating profile and the other with a more exploratory decision-making pattern. This approach sheds new light on the classic finding that all infants under 12 months tend to perseverate. Crucially, there were no significant differences in age between the clusters, but differences in terms of adaptive skills and temperament measured via gold-standard developmental assessments. These results prime the field for infant computational psychiatry, demonstrating that we can reliably fit models to infant data and that the parameters from such models can identify subgroups with distinct cognitive profiles that are superior to those derived from the behavioural data alone.

neuroscience↗

bioRxiv2024

Ultra-low-field brain MRI morphometry: test-retest reliability and correspondence to high-field MRI

Magnetic resonance imaging (MRI) enables non-invasive monitoring of healthy brain development and disease. Widely used higher field (>1.5 T) MRI systems are associated with high energy and infrastructure requirements, and high costs. Recent ultra-low-field (<0.1T) systems provide a more accessible and cost-effective alternative. However, it remains uncertain whether anatomical ultra-low-field neuroimaging can be used to reliably extract quantitative measures of brain morphometry, and to what extent such measures correspond to high-field MRI. Here we scanned 23 healthy adults aged 20-69 years on two identical 64 mT systems and a 3 T system, using T1w and T2w scans across a range of (64 mT) resolutions. We segmented brain images into 4 global tissue types and 98 local structures, and systematically evaluated between-scanner reliability of 64 mT morphometry and correspondence to 3 T measurements, using correlations of tissue volume and Dice spatial overlap of segmentations. We report high 64 mT reliability and correspondence to 3 T across 64 mT scan contrasts and resolutions, with highest performance shown by combining three T2w scans with low through-plane resolution into a single higher-resolution scan using multi-resolution registration. Larger structures show higher 64 mT reliability and correspondence to 3 T. Finally, we showcase the potential of ultra-low-field MRI for mapping neuroanatomical changes across the lifespan, and monitoring brain structures relevant to neurological disorders. Raw images are publicly available, enabling systematic validation of pre-processing and analysis approaches for ultra-low-field neuroimaging.

neuroscience↗

bioRxiv2024

Multi-orientation U-Net for Super-Resolution of Ultra-Low-Field Paediatric MRI

Owing to the high cost of modern MRI systems, their use in clinical care and neurodevelopmental research is limited to hospitals and universities in high income countries. Ultra-low-field systems with significantly lower scanning costs present a promising avenue towards global MRI accessibility, however their reduced SNR compared to 1.5 or 3T systems limits their applicability for research and clinical use. In this paper, we describe a deep learning-based super-resolution approach to generate high-resolution isotropic T2-weighted scans from low-resolution paediatric input scans. We train a multi-orientation U-Net, which uses multiple low-resolution anisotropic images acquired in orthogonal orientations to construct a super-resolved output. Our approach exhibits improved quality of outputs compared to current state-of-the-art methods for super-resolution of ultra-low-field scans in paediatric populations. Crucially for paediatric development, our approach improves reconstruction of deep brain structures with the greatest improvement in volume estimates of the caudate, where our model improves upon the state-of-the-art in: linear correlation (r = 0.94 vs 0.84 using existing methods), exact agreement (Lins concordance correlation = 0.94 vs 0.80) and mean error (0.05 cm3 vs 0.36 cm3). Our research serves as proof-of-principle of the viability of training deep-learning based super-resolution models for use in neurodevelopmental research and presents the first model trained exclusively on paired ultra-low-field and high-field data from infants.

neuroscience↗