Validation of FreeSurfer-Estimated Brain Cortical Thickness: Comparison with Histologic Measurements
Validation of FreeSurfer-Estimated Brain Cortical Thickness: Comparison with Histologic Measurements
复制标题
DOI：
10.1007/s12021-014-9229-2
复制
发表时间：
2014-10-01
期刊：
影响因子：
3
通讯作者：
Ferrigno, Giancarlo
中科院分区：
文献类型：
--
作者：
Cardinale, Francesco;Chinnici, Giuseppa;Ferrigno, Giancarlo
FreeSurfer软件包自动估算大脑皮质厚度。它的使用被广泛接受，尽管该工具仅在两次死后分离的脑磁共振扫描中通过组织学测量进行了验证。事实上，从未进行过组织学测量和FreeSurfer体内数据估计之间的比较。在“Claudio Munari”癫痫和帕金森手术中心，我们从2008年开始将FreeSurfer纳入我们的术前工作流程，主要是因为脑表面的自动重建有助于仔细规划手术切除。因此，我们将自动软件管道获得的皮质厚度值与从同一癫痫受试者的相应脑区切除的27个组织学标本进行的人工组织学测量进行了比较。这项方法比较研究，包括Passing-Bablok回归和Bland-Altman图分析，显示FreeSurfer估计值与皮层厚度的组织学测量值之间有很好的一致性。FreeSurfer和组织学测量的平均皮质厚度值（+/-标准差）分别为3.65 mm +/- 0.44和3.72 mm +/- 0.36 （P值= 0.32）。我们的发现加强了先前关于皮质厚度变化作为不同神经系统疾病生物标志物的报道。
FreeSurfer software package automatically estimates the cerebral cortical thickness. Its use is widely accepted, albeit this tool was validated against histologic measurements in only two post-mortem isolated brain MR scans. Indeed, a comparison between histologic measurements and FreeSurfer estimation from in vivo data was never performed. At the "Claudio Munari" Center for Epilepsy and Parkinson Surgery we have included FreeSurfer in our presurgical workflow since 2008, mainly because the automatic reconstruction of the brain surface is useful for carefully planning the surgical resection. We therefore compared cortical thickness values obtained by the automatic software pipeline with manual histologic measurements performed on 27 histologic specimens resected from the corresponding brain regions of the same epileptic subjects. This method-comparison study, including Passing-Bablok regression and Bland-Altman plot analysis, showed a good agreement between FreeSurfer estimation and histologic measurements of cortical thickness. The mean cortical thickness values (+/- Standard Deviation) obtained with FreeSurfer and histologic measurements were 3.65 mm +/- 0.44 and 3.72 mm +/- 0.36, respectively (P value = 0.32). Our findings strengthen previous reports on cortical thickness changes as biomarkers of different neurological conditions.