氟－18-FDG PET不同區域腦造影中List Mode影像之變異係數分析

Analysis of Coefficient of Variance in List Mode 18F-FDG PET Images of Various Brain Sub-regions

孫辰明(Chen-Ming Sun) ； 李浥涵(Yi-Han Li) ； 邱可莉(Ko-Li Chiou)

《核子醫學雜誌》 21卷4期 (2008/12) Pp. 205-210

https://doi.org/10.6332/ANMS.2104.002

摘要

背景：就標準差(standard deviation; SD)與均值(mean value)的比值之變異係數(coefficient of variance; CV)而言，變異性越小表示準確度(precision)越高。爲確保影像品質並減少檢查時間，本研究比較腦部之額葉、顳葉、枕葉及尾核等不同區域，利用SPM2(statistic parametric mapping)軟體分別統計出額葉、顳葉、枕葉及尾核箏不同區域上標準差及平均值，據此檢視可接受且無顯著差異性的理想收集時間。
方法：本研究針對31個正常腦進行氟－18-FDG PET造影，收集20分鐘的list mode影像，重組時間分別爲2分鐘、5分鐘、10分鐘及20分鐘四種。利用SPMZ進行頭部再定位(reorientation)，並分別進行空間正常化(spatial normalization)。爲求影像標準化，驗證每組影像內全部體素(voxel)之平均值(grand mean scaling)皆趨近1後，再做segmentation取其平均值之灰質機率影像。以此平均值之灰質(gray matter)機率影像再擷取其影像趨近灰質之80%機率，最後利用MRIcro version 1.40軟體分別在四個區域之mean image上以手動方式圈選ROI，分別紀錄ROI內所有體素值之標準差及其均值，求其變異係數。
結果：經由SPSS之成對樣本T檢定，閥值設在P=0.05，結果顯示在尾核(caudate)與顳葉(temporal lobe)的區域，2分鐘與20分鐘具顯著差異。顳葉(frontal lobe)則在不同時間皆有差異。枕葉(occipital lobe)部份，2分鐘、5分鐘皆與20分鐘有顯著差異。
結論：尾核與顳葉部份造影，避免使用5分鐘以下的收集時間。額葉部份造影，收集時間宜長。枕葉部份造影避免使用10分鐘以下的收集時間。

關鍵字

氟－18-FDG PET ； list mode影像收集模式 ； SPM statistic parametric mapping

並列摘要

Purpose: Coefficient of variance (CV), a relative measure of the variation, is defined as the ratio of the standard deviation (SD) to the mean. It is often referred to as the relative standard deviation (RSD or %RSD). This retrospective study aimed to find out if different emission duration time has influence on CV by using region of interest (ROI) method to compare CV of regional radiotracer distribution at frontal lobe, temporal lobe, occipital lobe and caudate.
Methods: Thirty-one healthy control subjects free of neurological diseases (age range, 40~74 years; mean±SD, 53.4±10.7 years) underwent 18F-FDG PET scans in list mode were included. The list mode image data in 4 group images (i.e., 2-min, 5-min, 10-min and 20-min) were reconstructed. All images were spatially normalized and proportional scaled using SPM2. A mean image was made from averaging all 120 images from 4 groups by the software Matlab through the image calculation interface of SPM2. ROI's were placed manually onto the mean image at bilateral caudate, frontal lobe, temporal lobe and occipital lobe, respectively, using the software MRIcro. The size of ROI of frontal, the temporal and the occipital lobe was 1,000 voxels, and ROI of caudate was 125 voxels. The SD, the mean and the CV of voxel values in each ROI were calculated. Paired t-test was applied to analyze the difference of CV for each ROI between 2, 5, 10 and 20-min images. The difference was statistically significant if P value ＜0.05.
Results: For caudate and temporal lobe, the CVs of 2-min images were significantly different from those of 20-min images, respectively. For frontal lobe, the CVs of 2-min, 5-min and 10-min images were significantly different from that of 20-min image. For occipital lobe, the CVs of 2-min and 5-min images were significantly different from that of 20-min images.
Conclusion: For imaging the regions of caudate and temporal lobe, the emission duration may not be less than 5 min. For imaging frontal lobe region, the suggested emission duration is as long as possible. For imaging occipital lobe region, the emission duration less than 10 min should be avoided. If variation of intra-subject regional radiotracer distribution is taken into account, we suggest that longer emission duration is more advisable.

並列關鍵字

18F-FDG PET ； list mode acquisition ； SPM statistic parametric mapping