複数の呼吸位相CT画像からの呼吸性変形量の算出とPETの画質改善への利用

摘要

PETは撮影に時間を要するため，胸部など呼吸により大きく動く部位に対しては画像にぼけが生じ，正確な診断を妨げる恐れがある．このため，10秒程度の息止め下でデータ収集・再構成を行い，これを同一呼吸相で複数回繰り返し，得られた画像群を加算する息止め撮影法が試みられている．しかし，息止めの繰り返し精度が悪い場合には，やはりぼけが生じる．これに対し我々は以前，PET画像群に対して互いの濃度分布が一致するように非線形に変形加算する方法を提案した．しかし，特に低濃度部分においては変形真の算出に十分な根拠を持たなかった．一方，近年のPETはCT装置と一体化されているものが多く，CTで得られた解剖学的画像データをPETの減衰補正やフュージョンに用いている．今回我々は呼気および吸気の息止めで撮影したCT画像から，呼吸による局所的な変形量を推定し，この情報を利用して息止めPET画像群の変形加算を行う方法を提案する．7例のデータに対して手法を適用した結果，6例のデ」タで腫瘍部位のコントラストが向上した．%Lung motion due to respiration causes image degradation in medical imaging, especially in nuclear medicine which requires long acquisition time. We have developed a method for image correction between the respiration-gated (RG) PET images in different respiration phases. The method offers the motion field vector of lung from the PET image set and the motion field vector plays a role for image summation of the RGPET images. This technique is applied to a set of breath-hold (BH) PET images in this paper. In the BH method, a patient is asked to hold his/her breath for 10 to 30 s as the image acquisition is performed. This is repeated several times and the obtained images are summed. In practice, however, a patient cannot necessarily hold his/her breath at the same timing of breathing. In such a case, the summed images still has a blur. We have proposed to apply our image registration method developed before for RG images to BH images. The BH images are nonlinearly motion-corrected so as to match a reference BH image and summed. In this paper, we introduced a new constraint on the motion vector field. Based on the use of PET/CT, we assumed to obtain two CT images; one is inhalation, the other is exhalation. From these images, we first calculated the motion vector field. In the registration of PET images, the motion correction of local point is limited to the direction given by those obtained by CT images. Through the experiment with six patient BH image data, we confirmed that the proposed method is effective in obtaining a non-blurred image.