产前超声深度学习模型识别胎儿脑室扩张
本研究基于超声自监督基础模型 USF-MAE，利用胎儿脑超声图像微调，实现正常与脑室扩张的二分类。模型在五折交叉验证和独立测试集上 F1 分数分别为 91.76% 和 91.78%，优于多种基线模型。结果显示在大型超声语料上预训练可提升分类性能和泛化能力。
要点
- USF-MAE 在测试集 F1 为 91.78%，对产前超声脑室扩张识别表现较强。
- 激活图显示模型在判断脑室扩张时主要聚焦于脑室区域。
- 该模型可作为产前胎儿脑结构评估的潜在辅助工具。
核心结论
- 在大规模超声专用语料上预训练可改善脑室扩张检测的分类性能与泛化能力。
- USF-MAE 框架具有较高的准确性、可靠性和可解释性，有望用于产前胎儿脑结构评估。
研究设计与样本
- 模型编码器在 OpenUS-46 语料库超过 370,000 张超声图像上完成预训练。
- 采用基于掩码自编码的超声自监督视觉 Transformer 基础模型 USF-MAE 进行微调。
- 通过五折交叉验证和独立测试队列评估模型性能。
- 基线模型包括 VGG-19、ResNet-50、ViT-B/16 和 MoCo v3。
关键影像表现/征象
- 激活图显示模型在识别脑室扩张时持续关注脑室区域。
诊断标准与关键数值
- 交叉验证集 F1 分数为 91.76%，测试集 F1 分数为 91.78%。
- 测试集平均精确率为 94.47%，准确率为 97.24%。
临床应用与阅片要点
- 可用于产前超声胎儿脑图像的脑室扩张辅助分类。
- 模型关注区域与脑室解剖位置一致，有助于提高判读信任度。
局限性
- 原文未报告样本量、中心数、孕周范围及参考标准等细节。
- 仅使用超声静态图像，未提及动态扫查或多切面信息。
以上解读依据论文摘要（全文未开放获取或许可不允许翻译转载）。
英文原文（摘要）· 点击展开
OBJECTIVE: To develop and evaluate a deep learning model capable of detecting ventriculomegaly on prenatal ultrasound images using a foundation model pre-trained specifically on ultrasound data.
METHODS: A vision transformer-based ultrasound self-supervised foundation model with masked autoencoding (USF-MAE) was fine-tuned for binary classification of fetal brain ultrasound images as normal or ventriculomegaly. The encoder had been pre-trained on more than 370,000 ultrasound images from the OpenUS-46 corpus. For this study, it was adapted and fine-tuned on a curated data set of fetal brain images. Model performance was evaluated through fivefold cross-validation as well as an independent test cohort. Accuracy, precision, recall, specificity, F1-score and area under the receiver operating characteristic and precision-recall curves were reported as the performance metrics. Eigen-CAM and Grad-CAM were used to visualize model attention.
RESULTS: USF-MAE reported an F1-score of 91.76% for the cross-validation set and 91.78% for the test set. The model outperformed all the baseline models, which included VGG-19, ResNet-50, ViT-B/16 and MoCo v3. The model reported a mean test precision of 94.47% and an accuracy of 97.24%. Activation maps indicated that the model consistently focused on the ventricular region when identifying ventriculomegaly.
CONCLUSION: Pre-training on a large, ultrasound-specific corpus improved classification performance and generalization for ventriculomegaly detection. The USF-MAE framework provided strong accuracy, reliability and explainability, showing potential as a robust tool for the assessment of fetal brain structures on prenatal ultrasound.
引用
- Megahed, Lee, Ducharme, et al. Deep Learning Analysis of Prenatal Ultrasound for Identification of Ventriculomegaly. Ultrasound Med Biol. 2026. PMID: 42236425. doi:10.1016/j.ultrasmedbio.2026.05.011
供执业医师工作参考，不替代临床判断；数值与分级以指南原文为准。