fNIRS, Parkinson’s disease, deep learning techniques, diagnostic accuracy
Abstract
Despite the difficulties associated with data processing, fNIRS technology shows promise for advancing cognitive studies and Parkinson’s disease research through the integration of deep learning techniques. The veracity and dependability of fNIRS data are contingent upon meticulous data collection, robust signal processing, and an acknowledgement of its inferior spatial resolution and restricted penetration depth in comparison to fMRI. The integration of resting and task state analyses using fNIRS provides a detailed insight into Parkinson’s disease, elucidating both the intrinsic brain connectivity disruptions and the dynamic responses to cognitive challenges. This enhances the diagnostic and treatment strategies employed in this field. The integration of fNIRS with EEG, motion capture, and advanced data analysis techniques markedly enhances the diagnostic accuracy of Parkinson’s disease. This is achieved by revealing distinct brain connectivity states and movement patterns, thereby paving the way for more sophisticated diagnostic and treatment approaches. The effective management of motion artefacts in fNIRS data for Parkinson’s disease research is achieved through the utilisation of advanced algorithms, including single-channel MAR, band-pass filtering and PCA. Collectively, these algorithms enhance the signal quality and facilitate the interpretability of brain activity patterns.
