MR-tagging画像における肝臓領域の非剛体変形量の自動計算手法に関する初期的検討

摘要

We propose an automatic calculation method of the degree of non-rigid deformity of the liver in MR Tagging images. In our method, bending energy was calculated, in which the thin-plate splines (TPS) is employed to deform an image using the movement of landmarks (LMs). Automatic tracing of crossing points of the grid lines in the MR tagging images is developed for searching the LMs. We used MR tagging images of 38 cases with 11 healthy livers and 27 chronic livers in evaluation. Bending energies obtained from the manual trace of the LMs and the automatic trace were compared. The significant difference in bending energies determined from the automatic traces was found between the cases of healthy liver and chronic liver.%本研究では，MR-tagging画像における肝臓領域の非剛体変形量の自動計算手法を提案する．提案手法は，（1）TPS（Thin-Plate Splines）法に基づく非剛体変形量の計算と，（2）非剛体変形量の計算に用いる格子状tagの交点の自動抽出からなる．格子状tagの手動抽出による交点と，自動抽出による交点のそれぞれの非剛体変形量の計算を，正常肝11症例と異常肝27症例からなる計38症例に対して行った．その結果，自動抽出による交点の非剛体変形量において，正常肝と異常肝の間に顕著な差が見られた．