Research and Development of quantitative CBF measurement with MRI

科学研究費補助金研究成果報告書研究種目: 基盤研究(C)研究期間: 2002～2004課題番号: 14571309研究代表者: 椎野 顯彦（滋賀医科大学・医学部・教授）研究分担者: 犬伏 俊郎（滋賀医科大学・MR医学総合研究センター・教授）これまでに報告してきた定量のための不確定因子とその対策法 1)AIFのpartial volume effectは、体軸方向に直角で大きな血管である、上矢状静脈洞(VOF)との比を求めることによって症例ごとに補正する。この結果、M1などの太い血管にこだわることなく、さらに細い動脈であっても同じ結果が得られることがわかり、M1の走行位置にとらわれずにスライス位置の選択が可能である。 2)Gradient echo (GE)法とSpin echo (SE)法。理論的にはSE法の方が太い血管に影響されにくくCBFの評価としては理想的であるが、虚血急性期の症例においては測定の失敗は極力避ける必要があるため、より感度のよいGE法が選択されることが多いと思われる。我々の施設では、multishot-EPIによりsusceptibilityの影響を少し下げるとともに、画像の歪みも抑え、良好な結果を得ている。 3)Pixel by pixel deconvolutionとSNR。Deconvolutionの際のnoiseによる定量値誤差の影響を調べるためには、ROI法を用いる。CBFを求めたのち、任意の部位に大きなROIを設定して、そのTICをdeconvolutionして求めたCBFの値とpixel-by pixelで求めた値と比較する。この時の比を求めればnoiseに由来する誤差を補正できる。 4)Noise filterの設定。P=0.2と固定するよりも、われわれの考案した方法p=1-average(s)/max(s)、を用いて、ピクセルごとに閾値を算出した方が、PETの定量値との相関係数が高く、優れていると思われる。 5)定量値の挿入として、白質を用いるのではなく、抽出してきたVOFのCBV値を104ml/100gとすることにより、白質病変による影響を避けることが可能である。また、VOFの抽出法によるばらつきも同時に補正することになり、より定量値の変動を抑えることが可能になる。Quantitative measurement of CBF by dynamic dynamic susceptibility contrast(DSC) technique is an important and challenging examination. For quantification of rCBF value on pixel-by-pixel basis, the arterial input function(AIF) has to be measured. The limited spatial resolution associated with the EPI sequences can potentially lead to inaccuracy and variability in the CBF measurements. In this study, a correction scheme is proposed to correct the errors. Subjects were studied with both MR and PET so that the accuracy of MR measured CBF can be assessed. Five healthy volunteers were studied. All MR images were acquired on a General Electric 1.5 Signa clinical scanner. Since single shot EPI sequence provide only a very low bandwidth in phase encoding direction leading to prominent signal misregistration, we used multishot EPI technique which minimized the signal loss or distortion. All images were transferred to a SGI Workstation for post-processing. The AIF and venous output function(VOF) were automatically obtained from large basal artery (IC or MCA or PCA) and the superior sagittal sinus, respectively. FT was employed to deconvolve the tissue function by the experimentally measured AIF and thus quantitative estimate of CBF was obtained. The AIF of each subject was corrected by the VOF, which was also obtained by integrating the delta-R_2^* curve obtained from the superior sagittal sinus. For normalization of the position between the MRI and the PET, three-dimensional anatomical fitting was performed. A total of 100 circular ROIs in each subject were placed in both the PET and MR CBF maps and were used to obtain regional estimate of CBF. As a result, there was good correlation in the CBF values among the two modalities (correlation coefficient=0.8) with the exception of the area including large vessels. Although several factors can potentially account for the observed variability in this study, the most plausible explanations are the partial volume effects of the AIF caused by the limited spatial resolution of the EPI sequence. The areas of AIF may be underestimated, leading to an overestimation as well as variability in the CBF measurement. We made an assumption that the integral of the delta-R_2^* curve of both the AIF and the VOF should be equal. The VOF corrected the partial volume effect and variables came from the AIF selection. In summary, we have presented an improved partial volume correction method, which reduces the errors in AIF determination