本論文利用掃描式白光干涉技術原理搭配快速傅立葉轉換法(Fast Fourier Transform, FFT)將時域干涉信號，藉由傅立葉振幅得到頻譜資訊，將其應用於局域性表面電漿共振效應(Localized Surface Plasmon Resonance, LSPR)之檢測，藉由LSPR吸收頻譜因環境折射率改變而產生紅移現象作為檢測依據。本文也利用質心演算法(Spectral Centroid Algorithm)來改善PZT(Piezoelectric Transducer)掃描長度限制頻譜解析度之缺點。 針對檢測系統可能造成的誤差來源，包含了波數精確性、光源照度不均勻之問題，進行研究與改善。波數精確性方面，利用一高通濾波片做為量測樣品，並將掃描式白光干涉儀與光譜儀(Hitachi U-4100)的量測結果作比較，當PZT移動一步的距離為71 nm時，與光譜儀量測之頻譜最為匹配，其平均差異為1.04 %；光源照度不均勻之問題，經由程式修正可將照度均勻度(Uniformity)約41.34 %改善至92.49 %。 在利用LSPR量測環境折射率變化方面，其檢測極限(Limit of Detection, LOD)可達1.2"×" 10-3 RIU (Refractive Index Unit)；應用於生醫檢測方面，本文使用rs242557 tau基因做為目標分析物，對其濃度之檢測極限為15 pM。 ;This article aims to set up a biosensor based on white-light scanning interferometry (WLSI) and applies to localized surface plasmon resonance (LSPR) sensor. In the WLSI system, the amplitude changes, as functions of wavelength, is retrieved by means of Fourier transform of the correlograms (interferograms in vertical direction). The Fourier amplitude method is capable to measure environmental changes in refractive index which results in a red shift of the amplitude (or spectrum) of gold nanoparticles (GNPs) solution. Meanwhile, the article shows that the performance of a spectrometer with poor spectral resolution is improved by the spectral centroid algorithm significantly because more data points (wavelengths) are considered for signal-to-noise ratio improvement Additionally, the sources of errors in the WLSI system, inclusive of the accuracy of wave-number and illumination non-uniformity, are improved. We use the spectrum of an edge filter as a check of the wavenumber accuracy. The result shows the best match with frame interval of 71 nm. In comparison to a spectrometer, the average spectrum error is 1.04 %. The illumination uniformity is also improved from 41.34 % to 92.49 % by the program correction. A method to measure the refractive index variation in the WLSI system is described in the article. The limit of detection (LOD) is 1.2"×" 10-3 RIU (Refractive Index Unit). The bio-applicability was investigated using the rs242557 tau gene, an Alzheimer′s and Parkinson’s disease biomarker. The LOD was calculated as 15 pM.