Maternal and Child Health Research Program, Department of OBGYN, University of Pennsylvania, Philadelphia, PA, USA.
Department of EECS, Vanderbilt University, Nashville, TN, USA.
J Ultrasound Med. 2022 Jun;41(6):1509-1524. doi: 10.1002/jum.15835. Epub 2021 Sep 23.
Early placental volume (PV) has been associated with small-for-gestational-age infants born under the 10th/5th centiles (SGA10/SGA5). Manual or semiautomated PV quantification from 3D ultrasound (3DUS) is time intensive, limiting its incorporation into clinical care. We devised a novel convolutional neural network (CNN) pipeline for fully automated placenta segmentation from 3DUS images, exploring the association between the calculated PV and SGA.
Volumes of 3DUS obtained from singleton pregnancies at 11-14 weeks' gestation were automatically segmented by our CNN pipeline trained and tested on 99/25 images, combining two 2D and one 3D models with downsampling/upsampling architecture. The PVs derived from the automated segmentations (PV ) were used to train multivariable logistic-regression classifiers for SGA10/SGA5. The test performance for predicting SGA was compared to PVs obtained via the semiautomated VOCAL (GE-Healthcare) method (PV ).
We included 442 subjects with 37 (8.4%) and 18 (4.1%) SGA10/SGA5 infants, respectively. Our segmentation pipeline achieved a mean Dice score of 0.88 on an independent test-set. Adjusted models including PV or PV were similarly predictive of SGA10 (area under curve [AUC]: PV = 0.780, PV = 0.768). The addition of PV to a clinical model without any PV included (AUC = 0.725) yielded statistically significant improvement in AUC (P < .05); whereas PV did not (P = .105). Moreover, when predicting SGA5, including the PV (0.897) brought statistically significant improvement over both the clinical model (0.839, P = .015) and the PV model (0.870, P = .039).
First trimester PV measurements derived from our CNN segmentation pipeline are significantly associated with future SGA. This fully automated tool enables the incorporation of including placental volumetric biometry into the bedside clinical evaluation as part of a multivariable prediction model for risk stratification and patient counseling.
早期胎盘体积（PV）与第 10/5 百分位数（SGA10/SGA5）以下的小于胎龄儿（SGA）有关。从三维超声（3DUS）手动或半自动定量 PV 非常耗时，限制了其在临床护理中的应用。我们设计了一种新的卷积神经网络（CNN）管道，用于从 3DUS 图像中全自动胎盘分割，探索计算出的 PV 与 SGA 之间的关系。
对 11-14 周妊娠的单胎妊娠的 3DUS 体积进行自动分割，我们的 CNN 管道通过 99/25 张图像进行训练和测试，结合了两个 2D 和一个 3D 模型，采用下采样/上采样架构。从自动分割中得出的 PV（PV）用于训练多变量逻辑回归分类器以预测 SGA10/SGA5。比较预测 SGA 的测试性能与通过半自动 VOCAL（GE-Healthcare）方法获得的 PV（PV）。
我们纳入了 442 名受试者，其中分别有 37（8.4%）和 18（4.1%）名 SGA10/SGA5 婴儿。我们的分割管道在独立测试集中的平均 Dice 评分为 0.88。包括 PV 或 PV 的调整模型对 SGA10 的预测结果相似（曲线下面积[AUC]：PV = 0.780，PV = 0.768）。将 PV 添加到没有任何 PV 包含的临床模型中（AUC = 0.725）可使 AUC 显著提高（P<.05）；而 PV 则没有（P=0.105）。此外，当预测 SGA5 时，包括 PV（0.897）与临床模型（0.839，P=0.015）和 PV 模型（0.870，P=0.039）相比，统计学上有显著改善。
从我们的 CNN 分割管道得出的早期妊娠 PV 测量值与未来的 SGA 显著相关。这种全自动工具使包括胎盘容积生物测量在内的胎盘体积生物测量能够纳入床边临床评估，作为多变量风险分层和患者咨询预测模型的一部分。
