Division of Nuclear Medicine, Department of Medical Imaging, Centre Hospitalier Universitaire Sainte-Justine, Montréal, Québec, Canada
Division of Nuclear Medicine, Department of Medicine, University of Ottawa and Ottawa Hospital, Ottawa, Ontario, Canada; and.
J Nucl Med. 2018 Jul;59(7):1118-1124. doi: 10.2967/jnumed.117.201889. Epub 2017 Dec 28.
Reference databases of pediatric brain metabolism are uncommon, because local brain metabolism evolves significantly with age throughout childhood, limiting their clinical applicability. The aim of this study was to develop mathematic models of regional relative brain metabolism using pediatric F-FDG PET with CT data of normal pediatric brains, accounting for sex and age. PET/CT brain acquisitions were obtained from 88 neurologically normal subjects, aged 6 mo to 18 y. Subjects were assigned to either a development group ( = 59) or a validation group ( = 29). For each subject, commercially available software was used to quantify the relative metabolism of 47 separate brain regions using whole-brain-normalized (WBN) and pons-normalized (PN) activity. The effects of age on regional relative brain metabolism were modeled using multiple linear and nonlinear mathematic equations, and the significance of sex was assessed using the Student test. Optimal models were selected using the Akaike information criterion. Mean predicted values and 95% prediction intervals were derived for all regions. Model predictions were compared with the validation dataset, and mean predicted error was calculated for all regions using both WBN and PN models. As a function of age, optimal models of regional relative brain metabolism were linear for 9 regions, quadratic for 13, cubic for 6, logarithmic for 12, power law for 7, and modified power law for 2 using WBN data and were linear for 9, quadratic for 25, cubic for 2, logarithmic for 6, and power law for 4 using PN data. Sex differences were found to be statistically significant only in the posterior cingulate cortex for the WBN data. Comparing our models with the validation group resulted in 94.3% of regions falling within the 95% prediction interval for WBN and 94.1% for PN. For all brain regions in the validation group, the error in prediction was 3% ± 0.96% using WBN data and 4.72% ± 1.25% when compared with the PN data ( < 0.0001). Pediatric brain metabolism is a complex function of age and sex. We have developed mathematic models of brain activity that allow for accurate prediction of regional pediatric brain metabolism.
小儿脑代谢的参考数据库并不常见，因为局部脑代谢在整个儿童期随年龄显著变化，限制了其临床应用。本研究的目的是利用小儿 F-FDG PET 与正常小儿脑 CT 数据，建立性别和年龄相关的区域性相对脑代谢数学模型。对 88 例神经正常的儿童（6 个月至 18 岁）进行了 PET/CT 脑采集。将受试者分为发育组（n=59）和验证组（n=29）。对于每个受试者，使用商业软件通过全脑归一化（WBN）和桥脑归一化（PN）活性来量化 47 个不同脑区的相对代谢。使用多元线性和非线性数学方程来模拟年龄对区域性相对脑代谢的影响，并使用学生 t 检验评估性别差异的显著性。使用赤池信息量准则选择最优模型。为所有区域推导了平均预测值和 95%预测区间。将模型预测与验证数据集进行比较，并使用 WBN 和 PN 模型计算所有区域的平均预测误差。作为年龄的函数，使用 WBN 数据，区域性相对脑代谢的最优模型为线性 9 个、二次 13 个、三次 6 个、对数 12 个、幂律 7 个、修正幂律 2 个，而使用 PN 数据则为线性 9 个、二次 25 个、三次 2 个、对数 6 个、幂律 4 个。WBN 数据中，仅在后扣带回皮质发现性别差异具有统计学意义。将我们的模型与验证组进行比较，WBN 的 94.3%区域和 PN 的 94.1%区域落在 95%预测区间内。对于验证组的所有脑区，WBN 数据的预测误差为 3%±0.96%，与 PN 数据相比为 4.72%±1.25%（<0.0001）。小儿脑代谢是年龄和性别的复杂函数。我们已经开发了大脑活动的数学模型，可以准确预测区域性小儿脑代谢。
