@misc{oai:repo.qst.go.jp:00072852,
 author = {田島, 英朗 and 赤松, 剛 and 岩男, 悠真 and 脇坂, 秀克 and 吉田, 英治 and 山下, 大地 and 山谷, 泰賀 and Tashima, Hideaki and Akamatsu, Go and Iwao, Yuma and Wakizaka, Hidekatsu and Yoshida, Eiji and Yamashita, Taichi and Yamaya, Taiga},
 month = {Jun},
 note = {Purpose: 
Dementia is a glowing social problem in the aging society. Toward early diagnosis of the dementia, we are developing a high sensitivity dedicated brain PET, helmet-neck PET, having a novel hemispherical detector geometry and an add-on detector at the back of the neck position. For PET systems, normalization is essential procedure to obtain accurate images. Rotating line sources or cylindrical phantom are typically used, but they cannot be placed inside the hemispherical gantry. Therefore, we have used a pool phantom fitting inside the detector geometry for the normalization of the helmet-neck PET. However, the attenuation medium of the pool phantom reduces the number of true event counts and increases scattered events, resulting in uncertainty of the normalization. In this study, we develop a hollow dome phantom as a more efficient normalization phantom without attenuation medium inside to reduce the attenuation and scattering. 
Material and Methods: 
The normalization phantoms were designed for the first helmet-neck prototype having 47 depth-of-interaction (DOI) detectors arranged in multiple rings with different diameters to form a hemispherical geometry with the inner diameter of 253 mm and 7 DOI detectors arranged as the add-on detector. We designed the hollow dome phantom with a combined shape of a hemisphere and a cylinder both with the inner diameter of 236 mm and the outer diameter of 248 mm having a cavity inside. The thicknesses of the walls and the cavity were 2 mm. The hollow phantom was fabricated by a photolithography technique. At first, we conducted Monte Carlo simulation using the Geant4 toolkit to compare the count rate of the true events and the scatter fraction of the pool phantom. Then we applied the developed hollow dome phantom to the helmet-neck prototype. 
Results:
The true count rates of the pool phantom and the hollow dome phantom at the radioactivity of 5 MBq were 17 cps/kBq and 83 cps/kBq, respectively. The true count rate of the hollow dome phantom was 4.8 times higher than the pool phantom. The scatter fractions were 31% for the pool phantom and 5.3% for the hollow dome phantom. We observed significant reduction of the scatter events.
Conclusions:
The hollow dome phantom is effective for normalization data acquisition for the helmet-neck PET because of high true count rate and small amount of scatter events., SNMMI 2018 Annual Meeting},
 title = {A new hollow-dome phantom for normalization of the helmet-neck PET},
 year = {2018}
}