Faculty
Social Science
Supervisor Name
Ingrid Johnsrude
Description
In order to surgically treat epilepsy, it is necessary to localize the epileptic lesion. Naturalistic functional magnetic resonance imaging (fMRI) can potentially be an accurate, non-invasive, and efficient tool for identifying diseased neural networks that cause epilepsy. We investigated inter-subject correlation (ISC) as a measure of neural synchronization between healthy controls (n = 24) and patients with epilepsy (n = 18) while subjects watched a stimulating movie clip. To investigate optimal denoising strategies, we analyzed ISC values with five sets of motion regressors, functions that remove unwanted motion-induced interference in fMRI signal. All ISC values were denoised using ICA-AROMA, an automated denoising technique with previously demonstrated efficacy. We found that denoising with ICA-AROMA and the set with the fewest amount of motion regressors produced higher correlation values, which supports this combination for use with naturalistic fMRI data. This was consistent at both the global and regional brain level. Further investigation of naturalistic fMRI would be useful to determine more strategies for optimal denoising as well as explore its clinical utility, which could greatly improve treatment and prognosis for patients with epilepsy.
Acknowledgements
I greatly thank Hana Abbas and Ingrid Johnsrude for their compassionate and unwavering guidance on this study. I extend my sincerest thanks and gratitude to all my co-authors at London Health Sciences Centre and Robarts Research Institute for their aid in developing the resources, materials, and data used in this study.
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.
Document Type
Poster
Included in
Optimal Denoising for Naturalistic fMRI Data
In order to surgically treat epilepsy, it is necessary to localize the epileptic lesion. Naturalistic functional magnetic resonance imaging (fMRI) can potentially be an accurate, non-invasive, and efficient tool for identifying diseased neural networks that cause epilepsy. We investigated inter-subject correlation (ISC) as a measure of neural synchronization between healthy controls (n = 24) and patients with epilepsy (n = 18) while subjects watched a stimulating movie clip. To investigate optimal denoising strategies, we analyzed ISC values with five sets of motion regressors, functions that remove unwanted motion-induced interference in fMRI signal. All ISC values were denoised using ICA-AROMA, an automated denoising technique with previously demonstrated efficacy. We found that denoising with ICA-AROMA and the set with the fewest amount of motion regressors produced higher correlation values, which supports this combination for use with naturalistic fMRI data. This was consistent at both the global and regional brain level. Further investigation of naturalistic fMRI would be useful to determine more strategies for optimal denoising as well as explore its clinical utility, which could greatly improve treatment and prognosis for patients with epilepsy.