Akademik Veri Yönetim Sistemi
Tezin Türü: Yüksek Lisans
Tezin Yürütüldüğü Kurum: Acıbadem Mehmet Ali Aydınlar Üniversitesi, Fen Bilimleri Enstitüsü, Tıp Mühendisliği Tezli Yüksek Lisans Programı, Türkiye
Tezin Onay Tarihi: 2020
Tezin Dili: İngilizce
Öğrenci: UMUT KARADENİZ
Danışman: Sinem Burcu Erdoğan
Özet:
Over the past two decades, fNIRS has become a practical and effective neuroimaging tool for investigating functional organization and specialization of the human brain. fNIRS systems have gained increasing popularity in neuroscience research due to non-invasive and light weight configuration and designs, field deployability, quick set-up time and planning. However, a major problem with fNIRS recordings has been the presence of various sources of noise mainly due to systemic physiological, experimental and instrumental artifacts. While instrumental noise can be eliminated by digital filters, a critical portion of the frequency spectrum of physiological effects caused by variations in heartbeat, blood pressure, respiratory activity and vascular tone overlap with that of the neuronally induced hemodynamic signals of interest. The presence of such systemic fluctuations degrades the signal to noise ratio by introducing false positive and false negatives besides real time interpretability of the signals. While many studies have emphasized the severity of physiological noise, no golden standard preprocessing technique could be proposed for eliminating systemic physiological artifacts which overlay hemodynamic signals induced by neuronal activation. Several methods for removal of systemic and motion artifacts from fNIRS signals have been proposed in the literature. However; no standardized pipeline of preprocessing steps could be reliably proposed as a gold standard for isolating the task related neuronally induced hemodynamic changes of interest. The aim of this study is to implement commonly applied noise reduction strategies in fNIRS literature into varying preprocessing pipelines and compare the performance of 48 different pipelines in increasing the signal to noise ratio in fNIRS signal collected during motor tasks.
Keywords: Functional near-infrared spectroscopy, hemodynamic response, preprocessing, denoising techniques, motion correction, band pass filtering, low frewquency noise removal