Breakey William, Knoops Paul G M, Borghi Alessandro, Rodriguez-Florez Naiara, Dunaway David J, Schievano Silvia, Jeelani Owase N U
*UCL Great Ormond Street Institute of Child Health †Craniofacial Unit, Great Ormond Street Hospital for Children, NHS Trust, London, UK.
J Craniofac Surg. 2017 Oct;28(7):1746-1751. doi: 10.1097/SCS.0000000000003929.
The ability to calculate intracranial volume (ICV) from 3-dimensional imaging is a useful tool in a craniofacial team's armamentarium. Intracranial volume uses range from decision making to assessment. Various methods to calculate ICV exist including fully manual, semiautomatic, and fully automatic techniques and they are used with varying frequency in craniofacial centres globally.This study aimed to systematically analyze and compare ICV calculations across the 3 methods and provide information to allow the reader to utilize these processes in practice.Twenty-six computed tomography scans from Apert patients were used to compare ICV measurements calculated using the following techniques: fully manual segmentation with OsiriX (taken as the gold standard); semiautomatic segmentation using Simpleware ScanIP; and fully automatic segmentation using FSL neuroimaging software. In addition, to assess the effect that a reducing CT scan slice number had on ICV measurement, 13 scans were remeasured using half, quarter, and an eighth of the slices of the full scan.The manual and semiautomatic techniques had intraclass correlation coefficients of 0.997, and 0.993 respectively. Intracranial volume measurements using the semi- and fully automatic techniques showed high linear correlation with manual techniques (R = 0.993 and R = 0.995). The coefficients of determination for full scan versus half, quarter, and eighth scan were R = 0.98, 0.96, and 0.94 respectively.Similar ICV results can be obtained using manual, semiautomatic, or automatic techniques with decreasing amount of time required to perform each method. Command line code for the fully automatic method is provided.
通过三维成像计算颅内体积（ICV）的能力是颅面团队工具库中的一项有用工具。颅内体积的用途涵盖从决策制定到评估等多个方面。存在多种计算ICV的方法，包括完全手动、半自动和全自动技术，并且它们在全球颅面中心的使用频率各不相同。本研究旨在系统分析和比较这三种方法的ICV计算结果，并提供相关信息，以便读者在实践中运用这些方法。使用来自Apert综合征患者的26份计算机断层扫描（CT）来比较使用以下技术计算的ICV测量值：使用OsiriX进行完全手动分割（将其视为金标准）；使用Simpleware ScanIP进行半自动分割；以及使用FSL神经成像软件进行全自动分割。此外，为了评估减少CT扫描层数对ICV测量的影响，对13份扫描分别使用全扫描层数的一半、四分之一和八分之一进行重新测量。手动和半自动技术的组内相关系数分别为0.997和0.993。使用半自动和全自动技术测量的颅内体积与手动技术显示出高度线性相关性（R = 0.993和R = 0.995）。全扫描与半扫描、四分之一扫描和八分之一扫描的决定系数分别为R = 0.98、0.96和0.94。使用手动、半自动或自动技术均可获得相似的ICV结果，且每种方法所需的时间逐渐减少。文中还提供了全自动方法的命令行代码。
