乳房 X 光攝影(X-ray mammography)長久用來作為乳房腫瘤診斷工具，由於結構性影像所限，至今仍存在相當之誤診率；而近紅外光擴散光學斷層掃描(Near Infrared Diffuse Optic Tomography, NIR DOT)則是近廿年開始發展之功能性腫瘤診斷技術。本研究旨在整合mammography 之結構影像、與 NIR DOT 之功能性影像，發展雙成像診斷技術。有效運用乳房X 光攝影之結構性影像，作為後續擴散光學斷層影像計算之初始猜值，如此將能達到快速收歛，及有效區分個別組織差異之目的。此研究基於過去數年在近紅外光擴散光學斷層掃描技術之基礎，包括直流(direct current)與頻域(frequency domain)之量測系統與組織光學係數(吸收係數及散射係數)反算程式，以及多個演算法分別用於快速收歛、保持差異組織邊緣以提昇演算效率等。本研究將發展並利用整合於乳房 X 光攝影機台壓板之外掛式NIR DOT 光學量測模組，進行資料擷取及後續組織光學係數影像反算，藉以獲得功能性斷層影像，驗證乳房腫瘤雙成像診斷系統之性能及成效，達到提昇乳房腫瘤診斷準確率的目的。X-ray mammography has been used to detect breast tumors for decades. There still exists high percentage of faulty diagnosis due to the limitation of structural information. Near infrared diffuse optic tomography (NIR DOT), a functional imaging modality, has been investigated and developed for twenty years. This study aims at developing a dual-modality imaging technique by incorporating the NIR DOT with the X-ray mammography. The structural images of mammogram are used as an initial guess of NIR DOT computation so that the tumor detection can achieve fast convergence and better performance for differentiate tumors from normal tissues. The research proceeds to the development of dual-modality tumor-detection technique based on previous basis, such as direct-current and frequency-domain measurement system, inverse computation scheme for optical-property images, and algorithms for the purpose of rapid convergence, edge-preserving for better tumor detection. The study is first to design an NIR-DOT measuring module incorporating with the press paddle of X-ray mammography. The module is used to acquire NIR data translating through tissues. While tumor detection is under process, X-ray mammograms and NIR data acquisition are taken alternatively. Subsequently, optical-property (absorption and scattering) images, functional images, can be inversely computed using the detected NIR data and the structural information of mammograms. In the study, the effectiveness of dual-modality diagnosis technique will be justified. To enhance the breast tumor detection is expected. 研究期間：10001 ~ 10012
