Effects of gradient coil noise and gradient coil replacement on the reproducibility of resting state networks

Epifanio Bagarinao, Erina Tsuzuki, Yukina Yoshida, Yohei Ozawa, Maki Kuzuya, Takashi Otani, Shuji Koyama, Haruo Isoda, Hirohisa Watanabe, Satoshi Maesawa, Shinji Naganawa, Gen Sobue

Research output: Contribution to journalArticle

Abstract

The stability of the MRI scanner throughout a given study is critical in minimizing hardware-induced variability in the acquired imaging data set. However, MRI scanners do malfunction at times, which could generate image artifacts and would require the replacement of a major component such as its gradient coil. In this article, we examined the effect of low intensity, randomly occurring hardware-related noise due to a faulty gradient coil on brain morphometric measures derived from T1-weighted images and resting state networks (RSNs) constructed from resting state functional MRI. We also introduced a method to detect and minimize the effect of the noise associated with a faulty gradient coil. Finally, we assessed the reproducibility of these morphometric measures and RSNs before and after gradient coil replacement. Our results showed that gradient coil noise, even at relatively low intensities, could introduce a large number of voxels exhibiting spurious significant connectivity changes in several RSNs. However, censoring the affected volumes during the analysis could minimize, if not completely eliminate, these spurious connectivity changes and could lead to reproducible RSNs even after gradient coil replacement.

Original languageEnglish
Article number148
JournalFrontiers in Human Neuroscience
Volume12
DOIs
Publication statusPublished - 19-04-2018

All Science Journal Classification (ASJC) codes

  • Neuropsychology and Physiological Psychology
  • Neurology
  • Psychiatry and Mental health
  • Biological Psychiatry
  • Behavioral Neuroscience

Fingerprint Dive into the research topics of 'Effects of gradient coil noise and gradient coil replacement on the reproducibility of resting state networks'. Together they form a unique fingerprint.

Cite this