近年來，隨著腦造影技術的進步，大腦的探索成為熱門的研究議題。尤其是非侵入式腦影像的發展，更進一步幫助人們了解腦部特定區域運作與其主要功能之間的關係。而近紅外光譜(Near-infrared spectroscopy，NIRS)是近年來受到注視的一種非侵入式腦影像技術，因為其具有靈敏的時間解析度及高空間解析度，並且能在開放空間中做動態實驗量測不同作業下的血流反應，因此受到許多研究團隊的採用。 本研究提出一套運用獨立成份分析法(Independent Componenet Analysis, ICA)於視覺誘發血流訊號的紅外光光譜分析方法，藉由獨立成份分析法，多通道紅外光血流訊號被分解成許多獨立成份(Independent component，IC)，每一個獨立成份接著與經由傳統平均方法所建立樣版模型訊號進行比對，選擇適當的獨立成份，進一步進行訊號重建，萃取出與視覺刺激相關的紅外光血流訊號。 本研究量測受視覺刺激誘發的紅外光訊號，並比較不同頻率和不同大小視覺刺激源的訊號。實驗結果證實，獨立成份分析法可有效的分離萃取出視覺訊號，進而提升訊號訊雜比。 With the advanced developments of brain imaging technologies, brain exploration has become a popular research field in recent years. Especially, the developments in noninvasive brain imaging techniques enable people to understand the functions of specific brain regions and the connectivities among different brain areas. Near infrared spectroscopy (NIRS) is a novel non-invasive brain imaging tool which measures hemodynamic response in accordance with subject’s specific performing tasks. It has the advantages of high temporal and high spatial resultions, and capability of being operated in open-space environment. Therefore, the NIRS has been chosen as an important tool by several research group for functional brain studies. In this study, we have proposed an independent component (ICA) – based analysis method for extracting visual-induced near infrared oxygen and deoxygen hemodynamic signals. With ICA, the multi-channel NIRS signals were first decomposed into a series of independent components (IC), and these ICs were correlated with a signal template, generated from traditional averaging method. Those ICs, which had high correlation values, were chosen as task-related components for constructing noise-suppressed hemodynamic response. This thesis studied visual-related near-infrared responses under visual stimuli with distinct sizes and frequencies. Our study results have shown the ICA is an effective tool to improve the signal-to-noise ratio (SNR) in visual-related NIRS signals.
