Division of Ultrasound in Obstetrics and Gynecology, Department of Obstetrics and Gynecology, New York University School of Medicine, NY 10016, USA.
Ultrasound Obstet Gynecol. 2010 Jul;36(1):20-5. doi: 10.1002/uog.7527.
To evaluate the feasibility and the processing time of offline analysis of three-dimensional (3D) brain volumes to perform a basic, as well as a detailed, targeted, fetal neurosonogram.
3D fetal brain volumes were obtained in 103 consecutive healthy fetuses that underwent routine anatomical survey at 20-23 postmenstrual weeks. Transabdominal gray-scale and power Doppler volumes of the fetal brain were acquired by one of three experienced sonographers (an average of seven volumes per fetus). Acquisition was first attempted in the sagittal and coronal planes. When the fetal position did not enable easy and rapid access to these planes, axial acquisition at the level of the biparietal diameter was performed. Offline analysis of each volume was performed by two of the authors in a blinded manner. A systematic technique of 'volume manipulation' was used to identify a list of 25 brain dimensions/structures comprising a complete basic evaluation, intracranial biometry and a detailed targeted fetal neurosonogram. The feasibility and reproducibility of obtaining diagnostic-quality images of the different structures was evaluated, and processing times were recorded, by the two examiners.
Diagnostic-quality visualization was feasible in all of the 25 structures, with an excellent visualization rate (85-100%) reported in 18 structures, a good visualization rate (69-97%) reported in five structures and a low visualization rate (38-54%) reported in two structures, by the two examiners. An average of 4.3 and 5.4 volumes were used to complete the examination by the two examiners, with a mean processing time of 7.2 and 8.8 minutes, respectively. The overall agreement rate for diagnostic visualization of the different brain structures between the two examiners was 89.9%, with a kappa coefficient of 0.5 (P < 0.001).
In experienced hands, offline analysis of 3D brain volumes is a reproducible modality that can identify all structures necessary to complete both a basic and a detailed second-trimester fetal neurosonogram.
评估离线分析三维（3D）脑体积以进行基本和详细的、有针对性的胎儿神经超声检查的可行性和处理时间。
对 103 例连续的健康胎儿进行了研究，这些胎儿在 20-23 孕周时接受了常规解剖学检查。由三位经验丰富的超声医师中的一位采集胎儿脑的经腹部灰阶和功率多普勒容积（每个胎儿平均采集 7 个容积）。首先尝试在矢状面和冠状面采集。当胎儿位置不利于快速获取这些切面时，在双顶径水平进行轴向采集。两位作者以盲法对每个容积进行离线分析。使用一种系统的“容积操作”技术来确定包括完整基本评估、颅内生物测量和详细的有针对性的胎儿神经超声检查在内的 25 个脑尺寸/结构的列表。两位检查者评估了获得不同结构的诊断质量图像的可行性和可重复性，并记录了处理时间。
两位检查者均报告 25 个结构中的所有结构均可进行诊断质量可视化，其中 18 个结构具有极好的可视化率（85-100%），5 个结构具有良好的可视化率（69-97%），2 个结构具有较低的可视化率（38-54%）。两位检查者平均使用 4.3 和 5.4 个容积完成检查，平均处理时间分别为 7.2 和 8.8 分钟。两位检查者对不同脑结构的诊断可视化的总体一致性率为 89.9%，kappa 系数为 0.5（P<0.001）。
在有经验的医生手中，3D 脑体积的离线分析是一种可重复的方法，可以识别完成基本和详细的中孕期胎儿神经超声检查所需的所有结构。
