Head motion correction based on pilot tone signals in MRI – a referenceless method

Motion remains a problem in clinical MRI, largely because all existing effective correction methods come with a penalty - constraints on pulse sequence parameters, expensive/bulky equipment, or extra steps in the workflow. The Pilot Tone (PT) is a small device that does not physically contact with the patient and only minimally impacts workflows. However, its signals can be difficult to process due to drifts, noise, and because they are difficult to model. We developed a motion correction algorithm that exploits the information from PT signals and tested the algorithm in 20 human subjects: four healthy volunteers and 16 patients. The study focused on a popular pulse sequence called 'magnetization prepared rapid gradient echo' (MPRAGE). A total of 156 MPRAGE scans were evaluated with quantitative and qualitative metrics. Images corrupted by artifacts were meaningfully improved by the proposed method, as demonstrated through increases in 'structural similarity index measure' (SSIM) and 'Dice similarity coefficient' (DSC) scores relative to ground truth scans and higher qualitative scores for motion corrected images compared to uncorrected images from three expert readers. Equally important, motion-free images were largely unaffected by the approach, as demonstrated through the expert reading scores and side-by-side image comparisons. The fact that the PT device requires no physical contact with the patient, coupled with the quantitative and qualitative effectiveness and relatively short reconstruction time (about 70 seconds per MPRAGE dataset), suggests that the approach could potentially succeed in transitioning to clinical use.