China Debarghya, MacLean Luke J, Wei Jinchi, Theodore Nicholas, Johnson Norbert, Crawford Neil, Ding Kai, Uneri Ali
Department of Biomedical Engineering, Johns Hopkins University, Baltimore, MD, United States.
Department of Neurosurgery, Johns Hopkins Hospital, Baltimore, MD, United States.
Med Image Anal. 2025 Oct;105:103672. doi: 10.1016/j.media.2025.103672. Epub 2025 May 31.
Recent advancements in machine learning (ML) allow for rapid analysis of complex image data, which supports the use of ultrasound (US)-based solutions in interventional procedures. These solutions often require large, labeled datasets that can be time-consuming to curate and subject to inter- and intra-labeler variability. This work presents a practical method for automated labeling of US images by transferring labels from 3D diagnostic images (e.g., CT or MR) using tracked US imaging to support supervised training. The approach was applied to segmenting spinal vertebrae, and the quality of the generated labels was evaluated by registering individual vertebrae from US to CT images to account for potential spinal deformation during surgery.
The proposed approach uses tracked US imaging to map target structures from CT volumes onto individual US frames. A dataset of spine images was created by scanning cadaveric torso specimens. Automated data cleaning methods were used to discard invalid frames, and data augmentations were applied to account for variability in image appearance. A simple U-Net model, called TernausNet, was trained for segmenting vertebrae using three labeling strategies: full vertebra (FV), posterior surface (PS), and weighted posterior surface (PS). The labels were evaluated through vertebrae segmentation and registration of the resulting segmentations to corresponding CT structures, considering the impact of labeling strategy, calibration errors, and data cleaning.
The proposed labeling strategies yielded improved segmentation accuracy over the direct mapping of CT labels (viz. FV), yielding a median of 5.18 [4.24, 6.66] mm RMSD for PS and 3.86 [2.87, 5.60] mm for PS labeling. The PS approach was particularly effective in reducing hallucination artifacts in the acoustic shadow regions below the vertebral cortex. Using the resulting segmentations, registrations were solved with 1.56 [1.30, 1.62] mm TRE for PS and 1.52 [1.32, 2.38] mm for PS labeling. Automated data cleaning and augmentation were found to significantly enhance the accuracy of bone feature segmentation and vertebra registration.
The study presents an automated labeling method for US imaging that supports the training of ML models by mapping 3D structures onto 2D US frames. The results highlight the importance of proper probe calibration, data cleaning, and specific labeling strategies in mitigating segmentation and registration errors. The work demonstrates the potential of real-time US imaging as a tool for precise anatomical tracking in surgery.
机器学习（ML）的最新进展使得能够对复杂图像数据进行快速分析，这为在介入手术中使用基于超声（US）的解决方案提供了支持。这些解决方案通常需要大量带标签的数据集，而精心整理这些数据集可能很耗时，并且会受到标注者之间和标注者内部差异的影响。这项工作提出了一种实用方法，通过使用跟踪式超声成像从3D诊断图像（如CT或MR）转移标签来自动标注超声图像，以支持监督训练。该方法被应用于分割脊椎，通过将超声图像中的单个椎体与CT图像配准来评估生成标签的质量，以考虑手术过程中潜在的脊柱变形。
所提出的方法使用跟踪式超声成像将CT容积中的目标结构映射到单个超声帧上。通过扫描尸体躯干标本创建了一个脊柱图像数据集。使用自动数据清理方法丢弃无效帧，并应用数据增强来考虑图像外观的变化。使用三种标注策略训练了一个名为TernausNet的简单U-Net模型来分割椎体：全椎体（FV）、后表面（PS）和加权后表面（PS）。通过椎体分割以及将所得分割结果与相应CT结构配准来评估标签，同时考虑标注策略、校准误差和数据清理的影响。
所提出的标注策略在分割精度上优于直接映射CT标签（即FV），PS标注的均方根偏差（RMSD）中位数为5.18 [4.24, 6.66] 毫米，PS标注为3.86 [2.87, 5.60] 毫米。PS方法在减少椎体皮质下方声影区域的幻觉伪影方面特别有效。使用所得分割结果，PS标注的配准平均误差（TRE）为1.56 [1.30, 1.62] 毫米，PS标注为1.52 [1.32, 2.38] 毫米。发现自动数据清理和增强显著提高了骨特征分割和椎体配准的准确性。
该研究提出了一种用于超声成像的自动标注方法，通过将3D结构映射到2D超声帧上来支持ML模型的训练。结果突出了正确的探头校准、数据清理和特定标注策略在减轻分割和配准误差方面的重要性。这项工作展示了实时超声成像作为手术中精确解剖跟踪工具的潜力。
