U. Pietrzyk et al., AN INTERACTIVE TECHNIQUE FOR 3-DIMENSIONAL IMAGE REGISTRATION - VALIDATION FOR PET, SPECT, MRI AND CT BRAIN STUDIES, The Journal of nuclear medicine, 35(12), 1994, pp. 2011-2018
A multipurpose three-dimensional registration technique was validated
with PET, SPECT, CT and MRI scans, which had been obtained under norma
l clinical conditions. In contrast to fully automated procedures, this
coregistration method is highly interactive, which has the advantage
that it does not impose rigid restrictions by data type and by alterat
ions in normal anatomy or brain function resulting from disease. Metho
ds: Basically, a computer program provides a variety of tools to exami
ne the accuracy of coregistration visually and to specify necessary tr
anslations and rotations in all three dimensions. Tools and criteria t
o accept coregistration were applied according to a standardized proto
col. Reproducibility was assessed with five independent users on nine
pairs of image sets. In two pairs of these image sets, coregistration
was repeated three times by each user. Results: Depending on the resol
ution of the images involved, the reproducibility of translation dista
nces ranged from 0.32 to 2.22 mm (s.d.) and of rotation angles from 0.
32 to 1.70 degrees. It was always much smaller than the point-spread f
ull-width half maximum of the device with the lower resolution. The ac
curacy of coregistration was examined using two arbitrarily misplaced
image sets. Interindividual and intraindividual variance were similar,
which suggested that the influence of subjectivity was not significan
t. Average displacements after coregistration were 0.43 and 0.29 mm or
less for PET and MRI data, respectively, which indicated the absence
of a systematic bias. Conclusion: The results indicate the high reprod
ucibility and accuracy of this three-dimensional coregistration techni
que, which is comparable or superior to those of automated techniques
and methods based on external artificial landmarks.