Skip to main content

Quantitative MRI of the brain using magnetic resonance fingerprinting

Periodic Reporting for period 1 - QuantMR7 (Quantitative MRI of the brain using magnetic resonance fingerprinting)

Reporting period: 2016-09-01 to 2018-08-31

Magnetic Resonance Imaging (MRI) is one of the most important diagnostic imaging techniques, with over 30 Million scans per year in the EU. Alterations in the intrinsic physical tissue parameters measured with MRI, such as of longitudinal (T1) and transverse (T2) relaxation times, have been implicated in major neurological conditions. Although these differences have been noted as useful signs in image contrasts, parameter quantification has not been exploited as a marker for disease stage or for monitoring treatment efficacy. Previous attempts to perform quantitative MRI protocols have suffered from sensitivity to system imperfections as well as infeasible, long acquisition times.

Recently, a new approach called MR fingerprinting (MRF) has been developed for the estimation of multiple parameters at once, featuring a new dedicated acquisition strategy. The method has shown great promise as data can be acquired in clinically acceptable time, is relatively insensitive to system imperfections and has high accuracy. In this project, we aimed to build on these novel techniques by demonstrating novel 3D MRF acquisition for imaging the whole brain using optimized acquisitions and reconstructions. We compared different strategies, using new automatic optimization algorithms and iterative reconstructions.

In improving imaging technology, research from QuantMR7 has worked towards addressing two significant limitations of current MRI examinations: firstly, the methods developed and demonstrated during the project go towards a quantitative rather than qualitative depiction of image information for radiological assessment and automated feature extraction; secondly, these methods allow for a simultaneous assessment of multiple parameters at once, greatly accelerating MRI examinations.
The work was mainly divided in three parts:

1) New k-space trajectories for acquisition have been implemented for 3D MRF scans and sequence parameters have been optimised.
In this part, acquisition parameters have been optimally selected for scanning the whole brain with a fast MRF scan. New algorithms based on Bayesian Optimization have been used in order to optimize acquisition schedules. Our flexible framework allowed the design of acquisition patterns tailored to specific experimental purposes, providing acquisition schemes for accurate quantitative maps.

2) New reconstruction algorithms and pulse sequences have been developed.
The acquisition sequences designed in this project were adapted for use in commercial MR scanners operating at three main static magnetic fields (1.5T 3T and 7T). Image reconstruction algorithms specific to the application were developed and tested, allowing faster acquisition and more reproducible and accurate quantifications.

3) Throughout the project the new techniques have been applied to healthy volunteers and patients.
We have acquired data in over 50 examinations. We have adapted existing software to analyse the quantitative data from whole-brain MRF data in volunteers at multiple scanners, in order to evaluate test-retest performances when using MRF in the healthy human brain at 1.5 T and 3.0 T. Test/retest Magnetic resonance fingerprinting was able to achieve an excellent repeatability and a good reproducibility in vivo in a short scan time. In addition, the techniques developed here were able to successfully discriminate damaged tissues in patients with different brain diseases.
We have demonstrated a novel protocol for quantitative whole-brain MRI, going from a developmental stage to the first applications. The application of the techniques developed in this project may have a major impact on the study and diagnosis of brain diseases. We have added on our sequence to research protocols being performed at the Stella Maris and IMAGO7 institutes, including neurological and developmental diseases. This has been useful to acquire preliminary data on over 50 patients. This data has been used to plan more structured clinical trials, which have now been funded and are starting from 2019.
In addition, our techniques have been shared with a number of other hospitals and universities, who have successfully replicated our results. After the required single-centre and multi-centre validation steps, the techniques developed here may ultimately benefit patients.