@misc{oai:repo.qst.go.jp:00070976,
 author = {Yoshikawa, Kyosan and Ohashi, Seiya and Toubaru, Sachiko and Hasebe, Mitsuhiko and Tamura, Katsumi and Shiraishi, Takahiro and Tanimoto, Katsuyuki and Yamada, Shigeru and Kandatsu, Susumu and Fukumura, Toshimitsu and Saga, Tsuneo and Kamada, Tadashi and 吉川 京燦 and 大橋 靖也 and 桃原 幸子 and 長谷部 充彦 and 田村 克巳 and 白石 貴博 and 谷本 克之 and 山田 滋 and 神立 進 and 福村 利光 and 佐賀 恒夫 and 鎌田 正},
 month = {Oct},
 note = {Purpose: Various parameters are used for measuring tumor glucose metabolic activity with 18F-FDG PET
including measuring the single maximum pixel value within the slice with highest radioactivity concentration
(SUVmax) or placing a fixed ROI in the area of the tumor with the highest level of 18F-FDG uptake (SUVpeak),
which it defines as the largest possible mean value of a 1 cm3 spherical VOI positioned within a tumor. SUVmax
is avoided inter-observer variation, but a major drawback of this approach is its susceptibility to noise in the
data. SUVpeak is also removed inter-observer variation, and the averaging of multiple voxels reduces the
susceptibility to noise. In this study we determined which parameter of glucose metabolic activity provided the
most robust predictions of patient prognosis of pancreas cancer treated by carbon ion radiotherapy.
Materials and Methods: A total of 41 patients (28 males, 13 females; mean age, 65 + 8+- y) with pancreas cancer
underwent 18F-FDG PET/CT scan before carbon ion radiotherapy. The following parameters on the
pretreatment scans were measured: maximum standardized uptake value (SUVmax), two types of peak SUV
(small: SUVpeak5, large: SUVpeak10), and average SUV (SUVave). SUVpeak5 and SUVpeak10 are measured
using spherical VOI with radius 5 mm and 10 mm, respectively. SUVave is an average pixel value of the circular
ROI with radius 5 mm within the slice with the highest radioactivity concentration. Correlation coefficients were
calculated for these parameters, and receiver operating characteristic (ROC) analysis was performed to
compare the prediction performance of each parameter for pancreas patients.
Results: Pretreatment SUVmax, SUVpeak5, SUVpeak10, and SUVave averaged 6.6, 6.1, 5.1, and 6.0,
respectively. Correlation coefficients between all four parameters, SUVmax and SUVpeak5, SUVmax and
SUVpeak10, SUVmax and SUVave, SUVpeak5 and SUVpeak10, SUVpeak5 and SUVave, SUVpeak10 and
SUVave, were 0.995, 0.964, 0.994, 0.973, 0.995, and 0.969, respectively. The areas under the ROC curve of
SUVmax, SUVpeak5, SUVpeak10, and SUVave were 0.69, 0.69, 0.92, and 0.70, respectively. ROC curves of
SUVmax, SUVpeak5 and SUVave were closely approximated to each other, but the ROC curve of SUVpeak10
was different from others and much better than others.
Conclusions: Four FDG-PET uptake parameters were correlated closely. ROC analysis showed that
SUVpeak10 was much better than others for prediction of pancreas patient prognosis. It seemed that relatively
large sphere for calculating SUVpeak is better than small sphere if the tumor is large enough for setting the
sphere within it., Annual Congress of the European Association of Nuclear Medicine},
 title = {Comparison of SUVmax and SUVpeak measuring FDG uptake for robust predictions of patient prognosis of pancreas cancer treated by carbon ion radiotherapy},
 year = {2012}
}