@inproceedings{rogers_fast_2022,
    author = {Rogers, D and Gao, Y and Boas, D A and Cronin-Golomb, A and Ellis, T D and Kiran, S and Somers, D C and von L{\"u}hmann, Alexander and Y{\"u}cel, Meryem},
    address = "Boston, USA",
    title = "Fast and slow movement-related artifacts in {fNIRS} signal: what is a viable solution?",
    copyright = "All rights reserved",
    abstract = "Methods: For this study, 12 subjects were prompted to act out motion artifacts that typically occur during human subject studies (tilting head left and right, tilting head up and down, lifting eyebrows, and speaking), over both a 5-minute rest period and another 5-minute rest period with 3 minutes of the rest containing multiple instances of the six previously mentioned motions. After data collection, 18s duration hemodynamic response functions (HRFs) were synthesized and overlayed on the motion artifact contaminated resting data. Then, the following correction methods were used to remove motion artifacts from the signal: No correction, SplineSG, PCARecurse, Wavelet3, tCCA with the correlation threshold parameter at .3 or the best correlating 3 regressors, and a combination of SplineSG and tCCA (at parameters 3 and .3). The group means, root mean square errors, and correlations were calculated and compared to the true/synthetic HRFs. Results: We found that though tCCA was effective in filtering slow physiological changes, it was not effective in filtering the high frequency distortions in the fNIRS signal. However, when combined with SplineSG, there was a larger improvement than each method separately. Conclusion: tCCA is an effective method in filtering systemic physiological changes; however, it is not effective in correcting for fast motion artifacts. We attribute the effectiveness of the SplineSG-tCCA combined method to the ability of SplineSG in correcting for fast motion artifacts and to the ability of tCCA in filtering the slow-motion artifacts. To compare the basis of all motion correction methods, none of the results used short separation regression. However, without using short separation regression, we did not take full advantage of the benefits of the tCCA method. In the future, we will repeat the study by adding short separation regression to all methods to determine the method that would both properly clean all motion artifact types and deal with the physiological contamination.",
    language = "en",
    booktitle = "Proc. {Biennial} {Meeting} of the {Society} for {fNIRS} 2022",
    publisher = "SfNIRS",
    month = "October",
    year = "2022",
    pages = "1"
}