Back to Projects

Image Acquisition and Reconstruction (pre-2019)

Fetal quantitative flow measurement by MRI

Project ID: 2017_102

Student: Caitlin Hardie

1st supervisor: Shaihan Malik, King’s College London
2nd supervisor: Jo Hajnal, King’s College London
3rd supervisor: Daniel Rueckert, Imperial College London

The aim of this project is to develop and validate q-flow methods for assessing the fetal cardiovascular system and cerebral circulation using an image domain approach with full motion correction based on kt-sampling, novel reconstruction and image registration methods.
The second aim of this project is to use the methods created to confirm normal flow ranges in a small cohort of normal fetuses and study clinical cases:

  1. with fetal cardiovascular malformations
  2. with fetal growth restriction

Magnetic Resonance Imaging (MRI) of the fetus can be highly challenging as a result of both fetal and maternal movements, and because of conflicting needs to both to image at high resolution and to encode a field of view large enough to encompass the maternal habitus. Our group and others have created a new and powerful capability to allow MRI to achieve fully motion corrected high resolution anatomic1,2,3, diffusion4,5  and functional6,7 fetal brain images, and we more recently created methods for whole fetus reconstruction from slice data8 and imaging the cardiovascular system (both intra cardiac cine imaging9 and extra cardiac anatomy10. All of these applications rely on rapidly imaging large numbers of individual slices that intersect the fetus, such that each slice is acquired fast enough to freeze the relevant fetal and maternal movements. Image realignment (image registration) is then used to subsequently transform from the scanner coordinate system into a coordinate frame that moves with the fetus. This creates a disordered dataset that can however be recovered onto a regular image matrix by appropriate interpolation. The fetal imaging group in Toronto, with whom we collaborate closely, have in parallel developed k-space methods for fetal cardiac11 and flow imaging12 that have enabled them to conduct important studies of cardiac function and vascular flow13.

In vivo quantitative flow measurement by MRI is conventionally performed using specially sensitised sequences in which flow velocity is encoded in the image phase (ref). Because image phase is also influenced by incidental extraneous factors such as local variations in the scanner’s main magnetic field, robust methods acquire both flow sensitised and reference phase images in an interleaved fashion, with the final result being obtained by subtraction. Proof that this can be achieved in fetuses that remain still for brief acquisitions has been demonstrated by the Toronto group12.

The aim of the project is to explore image domain methods for q-flow imaging that can accommodate the normal repertoire of fetal motion and so allow the approach to be applied routinely with a high success rate. Our work on image domain cine cardiac imaging (Von Amerom) suggests that transforming to the image domain before motion correction, extraction of the fetal heart rate and regeneration onto a regular image grid conveys a robustness and sensitivity that could be a major benefit for quantitative flow measurements. A key ingredient in this approach is imaging using acceleration methods that under-sample in k-space and time (so called kt methods) to achieve time series of images faster than the fetal heart rate (typically 150-180 beats per minute)

Back to Projects