RT Journal Article
SR Electronic
T1 Automated temporal subtraction scheme of torso FDG-PET scans by using a statistical shape model for normal cases
JF Journal of Nuclear Medicine
JO J Nucl Med
FD Society of Nuclear Medicine
SP 2097
OP 2097
VO 52
IS supplement 1
A1 Takeshi Hara
A1 Tetsuro Katafuchi
A1 Satoshi Ito
A1 Tokifumi Suzuki
A1 Takuya Matsumoto
A1 Xiangrong Zhou
A1 Hiroshi Fujita
YR 2011
UL http://jnm.snmjournals.org/content/52/supplement_1/2097.abstract
AB 2097 Objectives To develop an automated scheme of temporal subtraction for torso FDG-PET scan as a quantitative image analysis method by using a statistical method to obtain Z-scores of the lesion SUV. Methods We configured a normal model that indicates the normal range of SUV voxel by voxel. All of the normal scans were automatically registered into one model by using automated recognition method of bladder, liver, and body surface technique and 3D image deformation approach by using Affine transformation and thin-plate-spline technique. Z-scores of SUV (Score) were defined as deviations using the mean (M) and the standard deviation (SD) : Z-score(x, y, z) = {DeformedSUV(x, y, z) - M(x, y, z)} / SD(x, y, z). The mean and the standard deviation at a voxel were extracted from the normal model at the corresponding position in the deformed patient volume. The differences of SUV (ΔSUV) and Z-score (ΔZ-score) were obtained after the body region was deformed to fit the model. Results Normal cases of 243 (Male: 143, Female: 100) were employed to determine the normal metabolism distribution of FDG, and 63 abnormal cases were analyzed. The Z-scores for normal cases were in the range from zeros to plus/minus 2 SD. Most of the Z-scores in suspicious spots that the SUVmax were less than 5.0 showed statistical abnormalities (Z-score >2.0). Conclusions We have developed a quantitative image analysis scheme to estimate the abnormalities by the score of SUV by using the normal model and an automated subtraction method between current and previous PET scans. The subtraction images may be useful in interpretation of FDG-PET scans to confirm interval changes during chemotherapy and cancer progress diagnosis. Research Support This work is supported partly by a grant from Japan Government Grant-in-Aid for Scientific Research (C) 19500385