Objective: To report the feasibility of ICA on the BOLD signals of a Parkinson’s Disease patient.

Background: ICA is a mathematical method to decompose multivariate signals back into subcomponents to estimate the original sources of these signals. ICA analysis has recently entered the functional neurosurgery space in the last decade. Due to the current paucity in the literature regarding ICA on Parkinson’s patients, we present an analysis of a Parkinson’s case.

Method: An fMRI scan of a Parkinson’s Disease patient was analyzed. Traditional preprocessing steps such as motion correction, segmentation, high-pass filtering, time-slice correction, denoising and registration were initially performed to transform the fMRI into clearer, easier-to-work-with data. This was followed by a PCA to reduce the number of components. The FastICA algorithm was used to perform ICA on the data, and further noise was manually found and discarded. The components that matched known RSNs were found to make ICA maps.

Results: The ICA maps created visually represent the resting state networks of this patient. Components that correlate to resting state networks can be found both visually initially, and later by matching frequency data.

Conclusion: We report an ICA of a PD patient’s fMRI to be feasible. This is a novel technique that can help find and eventually compare the resting state networks of PD patients.

Figure 1

Figure 2

« Back to 2024 International Congress

MDS Abstracts - https://www.mdsabstracts.org/abstract/the-feasibility-of-ica-in-parkinsons-disease-fmri/