Lia M Hocke1,2, Blaise B Frederick1,2. 1. McLean Imaging Center, McLean Hospital, Belmont, Massachusetts, USA. 2. Department of Psychiatry, Harvard Medical School, Boston, Massachusetts, USA.
Abstract
PURPOSE: Motion estimation is an essential step in functional MRI (fMRI) preprocessing. Usually, fMRI processing software packages (eg, FSL and AFNI) automatically estimate motion parameters in order to counteract the effects of motion. However, the time courses of the motion estimation for fMRI data also contain information about physiological processes. Here, we show that respiration and cardiac signals can be extracted from motion estimation at significantly higher bandwidth than is possible with current methods. METHOD: To detect motion at high effective temporal resolution (HighRes), the motion parameters of stacks of simultaneously acquired slices were estimated separately, then combined. This method was validated by extracting physiological motion signals from resting state fMRI (rsfMRI) data (Enhanced Nathan Kline Institute-Rockland Sample) and comparing them to respiration belt and pulse oximeter signals. RESULTS: HighRes motion time-courses with an effective sampling rate of 15.5 and 11.4 Hz were extracted from repetition time (TR) = 0.645 and 1.4 s data, respectively. Respiration waveforms were extracted with significantly higher accuracy than the original motion parameters. Even cardiac waveforms could be extracted, despite the fact that the sampling time or TR values were too long to sample cardiac frequencies. CONCLUSION: HighRes motion traces provide insight into the subjects' motion at higher frequencies than can be estimated using standard techniques. In its simplest form, this technique can recover accurate respiration signals and may reveal additional complexity in brain motion.
PURPOSE: Motion estimation is an essential step in functional MRI (fMRI) preprocessing. Usually, fMRI processing software packages (eg, FSL and AFNI) automatically estimate motion parameters in order to counteract the effects of motion. However, the time courses of the motion estimation for fMRI data also contain information about physiological processes. Here, we show that respiration and cardiac signals can be extracted from motion estimation at significantly higher bandwidth than is possible with current methods. METHOD: To detect motion at high effective temporal resolution (HighRes), the motion parameters of stacks of simultaneously acquired slices were estimated separately, then combined. This method was validated by extracting physiological motion signals from resting state fMRI (rsfMRI) data (Enhanced Nathan Kline Institute-Rockland Sample) and comparing them to respiration belt and pulse oximeter signals. RESULTS: HighRes motion time-courses with an effective sampling rate of 15.5 and 11.4 Hz were extracted from repetition time (TR) = 0.645 and 1.4 s data, respectively. Respiration waveforms were extracted with significantly higher accuracy than the original motion parameters. Even cardiac waveforms could be extracted, despite the fact that the sampling time or TR values were too long to sample cardiac frequencies. CONCLUSION: HighRes motion traces provide insight into the subjects' motion at higher frequencies than can be estimated using standard techniques. In its simplest form, this technique can recover accurate respiration signals and may reveal additional complexity in brain motion.
Authors: Petra J van Houdt; Pauly P W Ossenblok; Paul A J M Boon; Frans S S Leijten; Demetrios N Velis; Cornelis J Stam; Jan C de Munck Journal: Hum Brain Mapp Date: 2010-02 Impact factor: 5.038
Authors: Steen Moeller; Essa Yacoub; Cheryl A Olman; Edward Auerbach; John Strupp; Noam Harel; Kâmil Uğurbil Journal: Magn Reson Med Date: 2010-05 Impact factor: 4.668
Authors: David A Feinberg; Steen Moeller; Stephen M Smith; Edward Auerbach; Sudhir Ramanna; Matthias Gunther; Matt F Glasser; Karla L Miller; Kamil Ugurbil; Essa Yacoub Journal: PLoS One Date: 2010-12-20 Impact factor: 3.240
Authors: C Hutton; O Josephs; J Stadler; E Featherstone; A Reid; O Speck; J Bernarding; N Weiskopf Journal: Neuroimage Date: 2011-04-15 Impact factor: 6.556