評估PET-CT融合影像之正確性

Evaluating Accuracy of Image Fusion for PET-CT

王寶英(Pao-Yin Wang) ； 陳泰賓(Tai-Been Chen) ； 劉佳宜(Chia-Yi Liu) ； 陳輝墉(Hue-Yong Chen)

《中華放射線技術學雜誌》 34卷4期 (2010/12) Pp. 258-263

https://doi.org/10.30045/CJRT.201012.0007

摘要

臨床上已普遍利用PET/CT融合影像進行病灶位置或病灶是否已跨區域瀰漫的判定。PET/CT是利用同床位且幾乎同時間收取的二組影像（PET與CT）分別進行影像重組再將二組影像融合，影像重建時正子影像(PET)與電腦斷層影像(CT)分別以不同矩陣大小進行；PET重建影像的像素矩陣可為168×168，256×256，或336×336像素；而CT重建影像則為512×512像素矩陣。由於二者硬體設備差異，使得重建影像時所設定的參數亦需調整。CT影像不僅可以提供解剖結構性資訊並同時提供PET影像進行衰減校正之用。然而重建影像時二組影像所設定的空間解析度參數不同將會影響影像融合後的對位偏差與失真。本研究對影像融合的評估，利用直線射源分別於正子影像與電腦斷層影像以不等比例（正子：168×168、336×336像素，CT: 512×512像素）及等比例像素（正子：256×256像素，CT: 512×512像素）進行影像重建，利用MATLAB對假體橫切面影像進行射源影像中心之座標差距比較，並利用曲線剖面圖(Line Profiles)之半高全寬評估正子影像的空間解析度。當PET像素矩陣為336×336時與CT為512×512之像素重建影像之融合影像不僅有良好的影像融合對位，而且提供更好的影像解析度。由實驗結果得知像素矩陣越大所得到的影像品質越佳，但在臨床作業時需一倂考量當矩陣加大時，所耗費的硬體空間與演算時間是否不致影響受檢者檢查之進行。良好的影像融合對位及影像空間解析度與矩陣大小具有相關性，建議CT與PET能使用相同之空間解析度與矩陣大小，如此才能消除對位偏差。

關鍵字

半高全寬 ； 正子與電腦斷層影像 ； 影像融合 ； 空間解析度

並列摘要

The fusion images of PET/CT are often used to diagnosis transformation and location of disease. The fusion images are combined PET and CT slices almost in the same time. We applied different reconstructed image spatial resolutions on clinical PET-CT scanner. The reconstructed image spatial resolutions are usually 512 x 512 for CT and 168 x 168, 256 x 256, or 336 x 336 for PET. The reconstructed conditions and parameters are different between PET and CT due to inconsistent hardware configurations. CT image not only support anatomical information for fusion with PET, but also provide CT-base attenuation correction for PET reconstructed image. The fusion images applied in clinical diagnosis will inconsistency and distortion due to heterogeneity of reconstructed conditions and image spatial resolutions from PET and CT. In this study, we evaluated the accuracy of fusion image between different image resolutions of PET and CT by using single line source at center of FOV. The reconstructed image resolutions were compared performances of unequal ratios (i.e., PET: 168 x 168 and 336 x 336, CT: 512 x 512) and equal ratio (i.e., PET: 256 x 256, CT: 512 x 512) between PET and CT. The horizontal line profile of phantom slice was performed to investigate the relation of peak location and reconstructed conditions. The full width at half maximum (FWHM) was adopted to evaluate spatial resolutions of PET image. The fusion image made by CT with size 512 x 512 and PET with size 336 x 336 was more accuracy and better spatial resolution then other combinations of PET and CT reconstructed conditions. The spatial resolution and image size affected the positioning of image fusion between CT and PET. In this study, we suggest that CT and PET use the same spatial resolution and image size to ensure positioning accuracy. The image quality was improvement as enlarging image matrix according to experimental results. However, the clinical diagnosis should take the computational cost of reconstruction into consideration as using high image resolution.

並列關鍵字

FWHM ； PET-CT ； Image Fusion ； Spatial Resolution