Department of Radiation Oncology, University of California San Francisco, United States of America.
Biomed Phys Eng Express. 2024 Aug 8;10(5). doi: 10.1088/2057-1976/ad6a63.
. Auto-segmentation in mouse micro-CT enhances the efficiency and consistency of preclinical experiments but often struggles with low-native-contrast and morphologically complex organs, such as the spleen, resulting in poor segmentation performance. While CT contrast agents can improve organ conspicuity, their use complicates experimental protocols and reduces feasibility. We developed a 3D Cycle Generative Adversarial Network (CycleGAN) incorporating anatomy-constrained U-Net models to leverage contrast-enhanced CT (CECT) insights to improve unenhanced native CT (NACT) segmentation.We employed a standard CycleGAN with an anatomical loss function to synthesize virtual CECT images from unpaired NACT scans at two different resolutions. Prior to training, two U-Nets were trained to automatically segment six major organs in NACT and CECT datasets, respectively. These pretrained 3D U-Nets were integrated during the CycleGAN training, segmenting synthetic images, and comparing them against ground truth annotations. The compound loss within the CycleGAN maintained anatomical fidelity. Full image processing was achieved for low-resolution datasets, while high-resolution datasets employed a patch-based method due to GPU memory constraints. Automated segmentation was applied to original NACT and synthetic CECT scans to evaluate CycleGAN performance using the Dice Similarity Coefficient (DSC) and the 95th percentile Hausdorff Distance (HD).High-resolution scans showed improved auto-segmentation, with an average DSC increase from 0.728 to 0.773 and a reduced HD95p from 1.19 mm to 0.94 mm. Low-resolution scans benefited more from synthetic contrast, showing a DSC increase from 0.586 to 0.682 and an HDreduction from 3.46 mm to 1.24 mm.Implementing CycleGAN to synthesize CECT scans substantially improved the visibility of the mouse spleen, leading to more precise auto-segmentation. This approach shows the potential in preclinical imaging studies where contrast agent use is impractical.
. 自动分割在小鼠 micro-CT 中提高了实验的效率和一致性，但在处理低原生对比度和形态复杂的器官，如脾脏时，往往会遇到困难，导致分割性能不佳。虽然 CT 造影剂可以提高器官的显著性，但它们的使用会使实验方案复杂化，降低可行性。我们开发了一种结合解剖结构约束的 U-Net 模型的 3D 循环生成对抗网络（CycleGAN），利用增强 CT（CECT）的见解来改善未经增强的原生 CT（NACT）分割。我们采用了标准的 CycleGAN 和解剖结构损失函数，从两个不同分辨率的未配对 NACT 扫描中合成虚拟 CECT 图像。在训练之前，我们分别训练了两个 U-Net 来自动分割 NACT 和 CECT 数据集的六个主要器官。这些预先训练的 3D U-Net 在 CycleGAN 训练过程中被整合，分割合成图像，并将其与地面真实标注进行比较。CycleGAN 中的复合损失保持了解剖结构的保真度。低分辨率数据集实现了全图像处理，而高分辨率数据集由于 GPU 内存限制，采用了基于补丁的方法。我们将自动分割应用于原始的 NACT 和合成的 CECT 扫描，使用 Dice 相似系数（DSC）和 95 百分位 Hausdorff 距离（HD）评估 CycleGAN 的性能。高分辨率扫描的自动分割得到了改善，平均 DSC 从 0.728 增加到 0.773，HD95p 从 1.19 毫米减少到 0.94 毫米。低分辨率扫描从合成对比中获益更多，DSC 从 0.586 增加到 0.682，HD 从 3.46 毫米减少到 1.24 毫米。通过实施 CycleGAN 来合成 CECT 扫描，大大提高了小鼠脾脏的可见度，从而实现更精确的自动分割。这种方法在对比剂使用不切实际的临床前成像研究中具有潜在的应用价值。
