How to obtain high-accuracy image registration: application to movement correction of dynamic positron emission tomography data

被引:38
作者
Andersson, JLR [1 ]
机构
[1] Univ Uppsala, PET Ctr, S-75185 Uppsala, Sweden
[2] Subfemtomole Biorecognit Project, Uppsala, Sweden
关键词
registration; dynamic; positron emission tomography;
D O I
10.1007/s002590050258
中图分类号
R8 [特种医学]; R445 [影像诊断学];
学科分类号
1002 ; 100207 ; 1009 ;
摘要
When registering dynamic positron emission tomography (PET) sequences, the time-dependent changes in uptake pattern prevent registration of all frames to the first frame in a straightforward manner. Instead, a sequential registration of each frame to its predecessor may be used, provided the registration algorithm is completely free of bias. It is shown that most existing algorithms introduce a bias, the size of which depends on the pixel size and the signal-to-noise ratio or the data. The bias is introduced by the pixelisation of the underlying continuous process. All existing cost-functions are more or less sensitive to noise, and the noise reduction resulting from translating one image set relative to the other means that a small movement will always be detected in the cases where no actual movement has occurred. The problem is solved by an initial resampling of the reference volume into a representation with another image and pixel size. If the new representation is sensibly chosen it means that all possible transforms applied to the other image volume will yield approximately the same noise reduction, thereby removing the source of the bias. The described effect is demonstrated on phantom data, and its impact is shown on human data.
引用
收藏
页码:575 / 586
页数:12
相关论文
共 26 条
[21]   Motion correction of PET images using multiple acquisition frames [J].
Picard, Y ;
Thompson, CJ .
IEEE TRANSACTIONS ON MEDICAL IMAGING, 1997, 16 (02) :137-144
[23]  
Snyder AZ., 1996, QUANTIFICATION BRAIN, P131, DOI DOI 10.1016/B978-012389760-2/50028-1
[24]   CONVOLUTION-BASED INTERPOLATION FOR FAST, HIGH-QUALITY ROTATION OF IMAGES [J].
UNSER, M ;
THEVENAZ, P ;
YAROSLAVSKY, L .
IEEE TRANSACTIONS ON IMAGE PROCESSING, 1995, 4 (10) :1371-1381
[25]   A NEW CLASS OF SIMILARITY MEASURES FOR ROBUST IMAGE REGISTRATION [J].
VENOT, A ;
LEBRUCHEC, JF ;
ROUCAYROL, JC .
COMPUTER VISION GRAPHICS AND IMAGE PROCESSING, 1984, 28 (02) :176-184
[26]   RAPID AUTOMATED ALGORITHM FOR ALIGNING AND RESLICING PET IMAGES [J].
WOODS, RP ;
CHERRY, SR ;
MAZZIOTTA, JC .
JOURNAL OF COMPUTER ASSISTED TOMOGRAPHY, 1992, 16 (04) :620-633