This paper focuses on the removal of periodic artifacts from neural signals recorded in rats in ultra-high field (UHF) MRI scanners, using a reference free adaptive feedforward method. Recording extracellular neural signals in the UHF environment is motivated by the desire to combine neural recording and UHF functional magnetic resonance imaging (fMRI) to better understand brain function. However, the neural signals are found to have extremely high noise artifacts of a periodic nature due to electromagnetic interference and due to small oscillatory motions. In particular, noise at 60 Hz and several harmonics of 60 Hz, sinusoidal noise from a pump, and low frequency breathing motion artifacts are observed. Due to significant overlap between the noise frequencies and the neural frequency region of interest, band pass filters cannot be effectively utilized in this application. Hence, this paper develops adaptive least squares feedforward cancellation filters to remove the periodic artifacts. The interference fundamental frequency is identified precisely using an implementation of k-means in an iterative approach. The paper includes significant animal data from rats recorded in an IACUC-approved procedure in 9.4 T and 16.4 T MRI machines. For breathing artifacts filtered from 4 rats, the mean signal cancellation values at the harmonic interference frequencies are 5.18, 12.97, and 20.87 dB/Hz for a sliding template subtraction, a single-stage impulse reference method, and the cascaded adaptive filtering approach respectively. For pump artifacts filtered from 2 chronically implanted rats, mean signal cancellation values are 2.85, 9.52 and 12.06 dB/Hz respectively. The experimental results show that periodic noise is very effectively removed by the developed cascaded adaptive least squares feedforward algorithm.
|Original language||English (US)|
|Journal||Biomedical Signal Processing and Control|
|State||Published - Jan 2022|
Bibliographical noteFunding Information:
This work was supported in part by NIH grant R01 MH111413 , the W.M. Keck Foundation , S10 RR025031 , NIBIB P41 EB027061 , P30 NS076408 , NSF IGERT 1069104 and by the University of Minnesota ’s MnDRIVE (Minnesota’s Discovery, Research and Innovation Economy) initiative.
© 2021 Elsevier Ltd
- Adaptive filters
- Artifact removal
- Breathing artifacts
- Extracellular neural signals
- MRI artifacts
- Motion artifacts
PubMed: MeSH publication types
- Journal Article