Article Text

Download PDFPDF

Advances in neonatal MRI of the brain: from research to practice
  1. Christopher J Kelly,
  2. Emer J Hughes,
  3. Mary A Rutherford,
  4. Serena J Counsell
  1. Centre for the Developing Brain, School of Imaging Sciences and Biomedical Engineering, King’s College London, London, UK
  1. Correspondence to Dr Christopher J Kelly, Centre for the Developing Brain, School of Imaging Sciences and Biomedical Engineering, King’s College London, London SE1 7EH, UK; christopher.kelly{at}

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

MRI is an important technique in neonatology that has the potential to improve care and parental counselling through better diagnosis and prognostication of disease. This review outlines the challenges of neonatal brain MRI, and explores current research in this field and the future translation of research techniques into clinical practice.


MRI is a relatively new and evolving technology—the first human scans were performed in the late 1970s,1 and the first neonatal brain scans followed shortly after in the early 1980s.2 Since then, the typical magnetic field strength used in clinical scanners has increased from approximately 0.15 Tesla to 1.5–3 Tesla, making higher resolution imaging possible, which is particularly important for the small anatomical structures in the neonatal brain. A variety of different MRI modalities are now possible, with both research and clinical applications (table 1).

Table 1

Types of MRI modality that are commonly used in neonatal brain imaging in clinical and research settings

Performing MRI scans in neonates can be technically challenging for a number of reasons: (1) size—the average brain volume of a term newborn is a quarter of an adult brain, requiring higher resolution to clearly delineate structures; (2) movement—motion corruption can lead to unusable images; babies are unable to follow instructions to lie still, and the combination of a noisy machine and unfamiliar environment can make it difficult to remain asleep; (3) physiology—higher resting heart rates and respiratory rates in neonates require careful adaptation of cardiac and flow-based imaging; and (4) biological differences—the immature brain has a higher water content and more unmyelinated white matter compared with adults, resulting in different tissue contrast from that of the adult brain and hence the need to optimise sequences accordingly.

To enable safe scanning of smaller and sicker neonates, a few research groups worldwide have colocated scanners directly on the neonatal intensive care unit, as early as 1996.3 Choice of scanner varies from adult magnets at higher field strengths (often 3T), to customised neonatal magnets with smaller bores and usually lower field strengths.3–5 Significant efforts have been made to streamline the imaging process, creating dedicated neonatal scanning systems to maximise chances of producing good-quality images suitable for full clinical interpretation.6–8

Technical research into neonatal brain MRI largely involves sequence optimisation, development of motion correction algorithms and development of techniques for subsequent image analysis. Close-fitting neonatal head coils improve the signal-to-noise ratio in neonatal brain imaging,6 9 allowing faster imaging at higher resolution. Motion tolerant MRI has the potential to revolutionise neonatal scanning.10 Traditionally, MRI is exquisitely motion-sensitive, often requiring oral sedation to achieve acceptable images in neonatal subjects. Recent techniques can tolerate a baby’s natural head movement within the scanner, obtaining structural images with no motion artefact11 (figure 1A,B).

Figure 1

Examples of neonatal brain MRI and analysis in the term newborn: (A) T2-weighted image with severe motion artefact, (B) same image as (A) but with motion correction algorithm applied, (C) T2-weighted image with diffusion tractography overlaid (left) and fibre orientation markers (right), (D) automatic segmentation of a T2-weighted image into 87 different tissue types, allowing precise volume measurements of the neonatal brain, (E) susceptibility-weighted image visualising venous structures and iron within the brain, and (F) non-contrast angiography.

Applications of neonatal brain MRI

Neonatal MRI is used clinically to assess acquired lesions such as hypoxic ischaemic injury seen with hypoxic ischaemic encephalopathy,12 13 haemorrhage,14 arterial stroke,15 infection16 and cases of non-accidental injury.17 18 Magnetic resonance (MR) can also improve the detection, visualisation and characterisation of congenital abnormalities, including vascular, cortical and white matter tract malformations, neural tube defects, and ventricular dilation. In the metabolic and genetic diagnostic workup, brain MRI can be helpful in providing additional information about structure, while MR spectroscopy allows measurements of metabolite concentrations and detection of specific abnormal peaks such as lactate or glycine in different regions of the brain.19

In the research setting, neonatal MRI is helping to improve our understanding of the normally developing brain, and the effects of prematurity and disease. Group studies are made possible through algorithms for automatic brain volume segmentation,20 cortical surface reconstruction21 22 and gyrification analysis.23 24 Diffusion tractography can delineate white matter tracts by determining the dominant direction of water diffusion at the microstructural level,25 and tract-specific analysis shows potential for analysing tissue differences within these tracts.26 The developing connectome of the brain can be explored using functional and diffusion MRI, leading to a better understanding of connectivity between regions in the neonatal brain.27

Translation of research into clinical practice

Despite recent advances in neonatal MRI research, it remains a hurdle to translate research into clinical practice. At the technical level, huge variation exists in scanning protocols and image quality around the UK and worldwide. This makes reproducibility and interpretation a challenge. The latest techniques from MRI research centres are slow to be translated to widespread usage. For example, motion-tolerant MRI currently requires custom scanner software, specialist technical input and powerful computing resources, and is not available for widespread use, despite the significant benefits that it offers. In order to make these newer techniques available to all neonatal units, we rely on scanner manufacturers to prioritise the translation of these promising research tools into standard features that run routinely on clinical scanners.

Performing MRI is time-consuming and resource-intensive. Therefore, clear evidence is required to demonstrate that imaging provides clinically relevant information that is not available by other less expensive methods, including clinical examination and readily available bedside ultrasound. In specific clinical settings, advanced techniques have become widely used. In cases of hypoxic-ischaemic encephalopathy13 28 29 and neonatal arterial stroke,30 MRI provides valuable prognostic information through visual interpretation of conventional sequences and through modalities including diffusion-weighted imaging and spectroscopy. Expertise in image interpretation is not widespread and referral of locally acquired images for central expert reporting may be necessary.

The benefit of routinely scanning preterm infants remains controversial, and there have been a significant number of studies investigating the effectiveness of using MRI to predict neurodevelopmental outcome in this population.31 Meta-analyses are unfortunately hampered by heterogeneous study populations, variable scanning techniques and differences in radiological analysis and interpretation. One systematic review concluded that MRI should preferably be performed at term-equivalent age,32 while another found that structural MRI before 36 weeks postmenstrual age has reasonable sensitivity and specificity to predict outcome of cerebral palsy.33 Very early scanning has been found to have limited value for predicting outcome in a preterm cohort born at a median of 27+4 weeks and imaged at just 2 days of life.34 However, MRI has still been found to be usefully performed well before term-equivalent age at a median of 32 weeks, with severity of abnormalities being associated with adverse early neurodevelopmental outcome.35 However, while larger studies have shown term-equivalent MRI to correlate with neurodevelopmental scores at school age,36 37 the 95% CIs (eg, for IQ) arguably remain too wide for effective longer term prognostic counselling with parents.38 A recent large study found that routine brain MRI in preterms reduced maternal anxiety slightly more than ultrasound, but that the effect was not statistically significant.39 MRI was also found to predict adverse motor outcomes slightly better than same-day ultrasound and neither predicted cognitive problems.39

Future directions

The ultimate goal of performing MRI in the neonatal period is to facilitate early diagnosis and appropriate management, while providing accurate and constructive prognostic information for parents. Many techniques outlined in this review remain limited to the research setting at present and are not widely available. Continued research is required to both improve neonatal MRI acquisition and interpretation, validate its effectiveness and prognostic power for clinical use, and make important newer techniques such as motion correction available for general use. New MRI sequences in development offer the promise of faster scan times, improved tissue contrast and better quantification of tissue properties.

As neonatal MRI becomes increasingly common, the training and development of experts in their interpretation will become increasingly important. While future developments in computer vision may assist automated interpretation of these complex images in the future, we need to both educate and train the next generation of radiologists to be prepared for the increasing usage of neonatal MRI.



  • Contributors CJK wrote the first draft. All authors contributed to the final draft.

  • Funding CJK is funded by a British Heart Foundation Clinical Research Training Fellowship (FS/15/55/31649). This work received funding from the Medical Research Council UK (MR/L011530/1), the European Research Council under the European Union’s Seventh Framework Programme (FP7/20072013)/ERC grant agreement no 319456 (dHCP project), and was supported by the Wellcome EPSRC Centre for Medical Engineering at King’s College London (WT 203148/Z/16/Z), MRC strategic grant MR/K006355/1, and by the National Institute for Health Research (NIHR) Biomedical Research Centre based at Guy’s and St Thomas' NHS Foundation Trust and King’s College London.

  • Disclaimer The views expressed are those of the authors and not necessarily those of the NHS, the NIHR or the Department of Health.

  • Competing interests None declared.

  • Provenance and peer review Not commissioned; externally peer reviewed.