WEKO3
アイテム
A Noise Reduction Method for Graphical Analyses with KL-Expansion during Transient Equilibrium Condition
https://repo.qst.go.jp/records/69791
https://repo.qst.go.jp/records/69791522de63d-a628-47fa-a941-e910cd8d984a
| Item type | 会議発表用資料 / Presentation(1) | |||||
|---|---|---|---|---|---|---|
| 公開日 | 2009-07-10 | |||||
| タイトル | ||||||
| タイトル | A Noise Reduction Method for Graphical Analyses with KL-Expansion during Transient Equilibrium Condition | |||||
| 言語 | ||||||
| 言語 | eng | |||||
| 資源タイプ | ||||||
| 資源タイプ識別子 | http://purl.org/coar/resource_type/c_c94f | |||||
| 資源タイプ | conference object | |||||
| アクセス権 | ||||||
| アクセス権 | metadata only access | |||||
| アクセス権URI | http://purl.org/coar/access_right/c_14cb | |||||
| 著者 |
Sakaguchi, Kazuya
× Sakaguchi, Kazuya× Naganawa, Mika× Sakata, Muneyuki× Shidahara, Miho× Seki, Chie× Ishiwata, Kiichi× Kimura, Yuichi× 坂口 和也× 長縄 美香× 坂田 宗之× 志田原 美保× 関 千江× 石渡 喜一× 木村 裕一 |
|||||
| 抄録 | ||||||
| 内容記述タイプ | Abstract | |||||
| 内容記述 | Objectives: Graphical analyses have advantages in unnecessary for the assumption of the number of compartments and fast computation. However, there is a large noise-induced bias or variability in the estimationof total volume of distribution (VT) [1]. Principal component analysis based method (mPCA) was proposed for a noise reduction in Logan graphical analysis (LGA) [2], which might lead to a deformation of kinetics of timeactivity curves (TACs). This study proposes a new noise reduction method for graphical analyses by Karhunen- Loève expansion (mKLE), which aims to maintain kinetics of voxel-based TACs. \nMethods: mKLE is based on the assumption that the ratio of a TAC to the input function is approximately constant after a certain time (t*) (transient equilibrium condition). Then the TAC after t* can be represented by one basis function. The basis function is expressed as an axis which passes across the origin and minimizes the distance from TACs in feature space. mKLE was applied to a set of TACs to calculate the basis function. The noise-reduced TACs were obtained by projecting voxel-based TACs onto the basis function. mKLE was tested on human PET data of [11C]TMSX, an adenosine A2A antagonist radioligand [3]. Dynamic PET scans were acquired for 1 hour, and metabolite-corrected arterial input function was obtained. We tested 3 versions: without noise reduction and with mKLE or mPCA. The dynamic data were analyzed by LGA and multilinear analysis (MA1) to calculate the parametric VT images. The regional mean value of VT was calculated from the striatum drawn on the summed PET image. For all graphical methods, t* was set to be 30 min postinjection. \nResults: Figure (A) shows a typical example of TAC from a voxel: an original TAC, the noise-reduced TACs by mKLE and mPCA, and the ROI-averaged TAC. The mKLE estimated TAC agreed well with the ROI-averaged TAC. Figure (B) shows the VT images from the same subject of Fig. (A). In the ROI-averaged TAC, the estimated VT values of LGA and MA1 were the same as 1.23 mL/cm3. The lowest value of VT was given by LGA without noise-reduction (1.12 +- 0.17 mL/cm3). Both mKLE and mPCA similarly improved VT values estimated with LGA (1.23 +- 0.19 mL/cm3 for mKLE, 1.24 +- 0.18 mL/cm3 for mPCA). On the other hand, only mKLE improved VT estimated with MA1 (1.24 +- 0.18 mL/cm3), while mPCA caused overestimation and large variability (1.37 +- 3.21 mL/cm3). \nConclusions: The experimental results suggest that the proposed method is promising for noise reduction of VT imaging by graphical analyses. [pic_01] \nReferences: [1] Slifstein, et al., J Nucl Med 41, 2083-2088, 2000. [2] Joshi, et al., J Cereb Blood Flow Metab 28, 852-865, 2008. [3] Ishiwata, et al., J Nucl Med, 41:345-354, 2000. |
|||||
| 会議概要（会議名, 開催地, 会期, 主催者等） | ||||||
| 内容記述タイプ | Other | |||||
| 内容記述 | Brain'09 & BrainPET'09 | |||||
| 発表年月日 | ||||||
| 日付 | 2009-07-03 | |||||
| 日付タイプ | Issued | |||||
