首页> 外文会议>IEEE Nuclear Science Symposium and Medical Imaging Conference >Fast atlas-based MRI-guided PET attenuation map generation in whole-body PET/MR imaging
【24h】

Fast atlas-based MRI-guided PET attenuation map generation in whole-body PET/MR imaging

机译:全身PET / MR成像中基于Atlas的快速MRI引导的PET衰减图生成

获取原文

摘要

We propose a one registration multi atlas (ORMA) pseudo-CT generation algorithm for attenuation correction in whole-body PET/MRI based on an optimized atlas-guided bone segmentation procedure. The proposed approach requires only one online registration between the target and reference images regardless of the number of atlas images N, while for the remaining subjects belonging to the atlas dataset, the pre-computed transformation matrices obtained from registration to the reference image is used to align them to the target image. The performance characteristics of the proposed method were evaluated and compared to conventional atlas-based attenuation map generation consisting of direct registration of the entire atlas images to the target. Four different PET attenuation maps were produced using direct registration and ORMA, voxelwise weighted (VWW-Direct and VWW-ORMA) and arithmetic average (AA-Direct and AA-ORMA) atlas fusion strategies. The comparison of validation measures characterizing the accuracy of extracted whole-body bone demonstrated the superiority of VWW-Direct atlas fusion technique resulting in a Dice similarity measure of 0.82±0.04 compared to 0.60±0.02 for AA-Direct. Conversely, the ORMA approach yielded a Dice similarity measure of 0.76±0.05 for VWW-ORMA and 0.55±0.03 for AA-ORMA. Quantitative analysis of PET data revealed good correlation (y=1.01x+0.1, R2=0.99) between PET images corrected for attenuation using the proposed pseudo-CT method and the corresponding reference CT images. The proposed method generates decent attenuation maps and enables to reduce the processing time for atlas-based pseudo-CT generation reduced by a factor N.
机译:我们基于优化的图集引导的骨分割程序,提出了一种用于全身PET / MRI衰减校正的单配准多图集(ORMA)伪CT生成算法。所提出的方法仅需要在目标图像和参考图像之间进行一次在线配准,而与地图集图像N的数量无关,而对于属于地图集数据集的其余主体,则使用从配准到参考图像获得的预先计算的变换矩阵来进行配准。将它们与目标图像对齐。评估了所提出方法的性能特征,并将其与传统的基于地图集的衰减图生成(包括将整个地图集图像直接配准到目标)进行了比较。使用直接配准和ORMA,体素加权(VWW-Direct和VWW-ORMA)和算术平均值(AA-Direct和AA-ORMA)图集融合策略制作了四个不同的PET衰减图。表征提取的全身骨骼准确性的验证措施的比较证明了VWW-Direct Atlas融合技术的优越性,与A-Direct的0.60±0.02相比,Dice相似性度量为0.82±0.04。相反,对于VWW-ORMA,ORMA方法产生的Dice相似度度量为0.76±0.05,而对于AA-ORMA,则为0.55±0.03。 PET数据的定量分析显示,使用拟定的伪CT方法校正后的PET图像与相应的参考CT图像之间具有良好的相关性(y = 1.01x + 0.1,R2 = 0.99)。所提出的方法生成体面的衰减图,并能够将基于图集的伪CT生成的处理时间减少N倍。

著录项

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱:kefu@zhangqiaokeyan.com

京公网安备:11010802029741号 ICP备案号:京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有
  • 客服微信

  • 服务号