Back to Projects

AI-enabled Imaging, Emerging Imaging

Motion corrected neuroimaging using multi-frequency pilot tone 

Project ID: 2022_020

1st Supervisor: Joseph Hajnal, King’s College London
2nd Supervisor: Shaihan Malik, King’s College London
Clinical Supervisor: Prof Alexander Hammers, King’s College London

Aim of the PhD Project:

  • Develop fast and robust motion prediction for neuroimaging using ML with minimal training based on single frequency pilot tones 
  • Combine higher frequency tone information for surface detection to maximise motion sensitivity 
  • Prospective motion correction with high precision using advanced motion models

Project description/background:

Patient motion is a perennial challenge for high resolution MR imaging, and often leads to wasted resources since acquisitions may need to be repeated as artefacts destroy their diagnostic value. A wide range of motion correction approaches exist in the literature, including both prospective and retrospective methods, using both external sensors and/or the MR data itself. For neuroimaging, correction of (rigid) head motion can be accurately achieved by use of external optical sensors1 to track the position of the subject’s head. This data can be used to correct the image field of view prospectively (i.e. during acquisition) or correct the reconstruction retrospectively. Unfortunately, to track head motion with sub-millimetre precision, optical sensors must be combined with marker(s) physically fixed to the skull – this is achieved for example using a marker held between the subject’s teeth. While very accurate, patients may not tolerate it. Meanwhile, data-based retrospective motion correction can often achieve excellent results2 but are limited in applicability to certain pulse sequences, and may require additional data to be collected, making scanning less efficient.  

In a recent development, a novel method for detecting patient respiration was introduced3. A constant frequency radio signal (i.e. a ‘pilot tone’) is played in the scanner room, at a frequency that will be directly picked up by the MRI receivers. The frequency of this tone is set so that it can be detected by the scanner, but it will not cause image artefacts. As the patient respires, the ‘loading’ of the MRI receiver coil will periodically change since the composition of tissue immediately adjacent to it changes. The resulting change in coil impedance causes the relative amplitude of the ‘tone’ as detected by this coil to also fluctuate. By monitoring the detected pilot tone, it is possible to infer the position in the respiratory cycle of the patient.  

Since neuroimaging uses dense arrays of small receivers placed close to the subject’s head, small changes in head position will differentially change the loading of each detector. Hence detection of head motion should be possible by the same principle. We have recently demonstrated this on a 7T scanner, and have been able to infer translation and rotation parameters of a subject’s head by monitoring a pilot tone signal as a proof of concept4. The motion information gained could ultimately be used to improve the quality of prospective or retrospective motion correction methods, and could have wide applicability for all forms of MRI neuroimaging potentially at a range of scanner field strengths.  

This project will seek to develop an accurate motion inference approach by using multi-frequency pilot tones and neural networks for fast prediction of motion parameters. The project will include assembly of new hardware, imaging experiments and data analysis, and would suit a candidate with a strong physics or engineering background. 

References

  1. Callaghan, M. F.et al.An evaluation of prospective motion correction (PMC) for high resolution quantitative MRI. Front. Neurosci. 9, 1–9 (2015). 
  2. Cordero-Grande, L.et al.Motion-corrected MRI with DISORDER: Distributed and incoherent sample orders for reconstruction deblurring using encoding redundancy. Magn. Reson. Med. 1–14 (2020) doi:10.1002/mrm.28157. 
  3. Ludwig, J., Speier, P., Seifert, F., Schaeffter, T. & Kolbitsch, C. Pilot tone–based motion correction for prospective respiratory compensated cardiac cine MRI.Magn. Reson. Med.85, 2403–2416 (2021). 
  4. Wilkinson, T.et al.Motion Estimation for Brain Imaging at Ultra-High Field Using Pilot-Tone: Comparison with DISORDER Motion Compensation. in Proc Intl Soc Mag Reson Med 2021 122 (2021). 
  5. Anand, S. & Lustig, M. Beat Pilot Tone: Exploiting Preamplifier Intermodulation of UHF/SHFRF for Improved Motion Sensitivity over Pilot Tone Navigators. inProc Intl Soc Mag Reson Med 2021 568 (2021). 
  6. Jaeschke, S. H. F., Robson, M. D. & Hess, A. T. Scattering matrix imaging pulse design for real‐time respiration and cardiac motion monitoring.Magn. Reson. Med.2169–2177 (2019) doi:10.1002/mrm.27884. 

Back to Projects