2010-08-012024-05-15https://scholars.lib.ntu.edu.tw/handle/123456789/666223摘要：腦磁圖(magnetoencephalography, MEG)使用高敏感度的超導量子干涉元件(SQUID)來量 測由腦神經元在動作時所產生之微弱磁場. 藉由解”逆問題(inverse problem)” 我們 可以由頭外量測的MEG信號來反推人腦在工作時神經電流源的位置, 方向, 以及大小, 藉此以解開人腦在工作時複雜的交互作用.最小電流估計法(minimum current estimate, MCE)是使用MEG資料估計神經電流源分布的一種方法, 其特色在於能整合高 空間解析度的解剖資訊(由核磁共振影像MRI提供). 且估計電流源分布時不需要事先指 定電流源的數目, 估計出的結果因為數學上最小L1範數(minimum L1 norm)的限制使的 電流元的空間分布較為集中, 相對於傳統分散式電流源分布(distributed source modeling; 例如最小範數估計minimum-norm estimate)的空間解析度較高. 然而MCE的 缺點在於它的解對於雜訊的敏感度太高, 以至於對於模型中些微的誤差都會造成估計電 流源在時間以及空間上的強烈不連續性. 有鑒於此, 我們希望研究出新的基於最小L1範 數的分散式電流源分布法.此方法稱為” 穩健腦磁圖訊號源偵測法” (RoFos), 他可以 自動整合由MRI所提供的解剖資訊來建議可能的電流源方向.並且改變在傳統MCE最佳化 時的限制, 自動有彈性的調整模型描述的誤差以及估計電流源的L1 norm大小, 以達到 在兼顧兩者之下最低成本(cost)的目的. RoFos將可在數值模擬以及活體實驗資料中展 示出高空間解析度以及高時間空間穩定度的電流源估計值. 此外, 我們另將探討RoFos 的統計分布, 藉此將估計出的電流元大小轉換成動態統計參數圖(dynamic statistical parametric map, sRoFos). 發展出的RoFos 以及 sRoFos 將可應用於高時間空間解析 度的人腦功能映像上. <br> Abstract: Magnetoencephalography (MEG) non-invasively detects weak extracranial magnetic fields during tasks using super-conducting quantum interference devices (SQUIDs). The MEG inverse problem is to quantitatively model the underlying neuronal current sources by estimating their locations and orientations using MEG measurements. One method of MEG source modeling is the minimum-current estimate (MCE), which provides spatially focal source distribution. The knowledge of current source orientations required in MCE can be first estimated by the minimum-norm estimate (MNE) with the loose orientation constraint (LOC). However, MCE is unstable because it is sensitive to noise and it inherits any errors due to the inaccurate source orientations estimated by the LOC. Similar to the MCE minimizing the l1-norm of the source model, here we study the robust focal source modeling (RoFoS), which can automatically integrates the current source orientation information into a cost function to jointly model the residual error and estimated source strengths. We attempt to show that RoFoS has a smaller spatial extent of source estimates and higher localization accuracy using simulations and in vivo MEG experimental data. Additionally, we will numerically validate the null distribution of RoFoS and subsequently derive the dynamic statistical parametric inferences (sRoFoS). The RoFoS and sRoFoS methods can be used for high spatiotemporal resolution imaging of human brain function measured by MEG and possibly EEG.腦磁圖超導量子干涉元件逆問題穩健腦磁圖訊號源偵測法MEGSQUIDsRoFoS