文献速递核素与放射性配体治疗肿瘤 PET/CT

基于深度学习的无CT衰减校正与蒙特卡洛散射校正引导的定量90Y-SPECT成像以改进剂量计算

CT-free attenuation and Monte-Carlo based scatter correction-guided quantitative 90Y-SPECT imaging for improved dose calculation using deep learning.
PubMed 文献 · Eur J Nucl Med Mol Imaging · 2025-03-13

本研究纳入190例接受玻璃微球90Y选择性内放射治疗(SIRT)的患者,训练三种深度学习模型分别完成衰减校正(AC)、散射校正(SC)及联合校正(ASC),在未见测试集上比较器官和体素级剂量学结果与参考剂量图的差异,并以伽马分析探索适合SPECT SIRT剂量学的标准。

要点

  • 针对AC、SC、ASC三种任务分别训练基于Swin UNETR架构的深度学习模型,训练采用五折交叉验证。
  • 体素级伽马分析在三种标准下通过率均约98%,肿瘤平均绝对误差高于全正常肝脏。
  • 作者认为这些任务特异性模型在CT不可用或不可靠时可辅助完成必要校正,需在更大数据集训练后应用。

核心结论

  • 作者利用经患者特异性蒙特卡洛散射校正和基于CT的衰减校正图像,针对AC、SC、ASC三种临床场景开发了多个模型。
  • 作者认为这些任务特异性模型在CT图像不可用或因配准不良不可靠时,经更大数据集训练后可能有助于完成必要的校正。

研究设计与样本

  • 研究纳入190例接受玻璃微球90Y选择性内放射治疗(SIRT)的患者数据。
  • 在未校正和校正SPECT图像上采用局部能量沉积法进行体素级剂量学计算。
  • 分别针对AC、SC和联合ASC训练三个深度学习模型,采用改进的3D移位窗口UNet Transformer(Swin UNETR)架构。
  • 校正后剂量图作为参考、未校正剂量图作为输入,数据按约80%训练集和约20%未见测试集划分,训练采用五折交叉验证。

主要发现

  • AC任务体素级定量指标均值±标准差:平均误差(ME)为-0.026±0.06 Gy,结构相似性指数(SSIM)为99.5±0.25%,峰值信噪比(PSNR)为47.28±3.31 dB。
  • SC任务上述指标分别为-0.014±0.05 Gy、99.88±0.099%、55.9±4 dB。
  • ASC任务上述指标分别为-0.04±0.06 Gy、99.57±0.33%、47.97±3.6 dB。
  • 三种体素级伽马分析标准(DTA 4.79 mm/DD 1%、DTA 10 mm/DD 5%、DTA 15 mm/DD 10%)的通过率均约为98%。
  • 肿瘤与全正常肝脏的平均绝对误差(MAE):AC为7.22±5.9 Gy与1.09±0.86 Gy,SC为8±9.3 Gy与0.9±0.8 Gy,ASC为11.8±12.02 Gy与1.3±0.98 Gy。

关键数值与推荐

  • 体素级伽马分析采用三种标准:DTA 4.79 mm/DD 1%、DTA 10 mm/DD 5%、DTA 15 mm/DD 10%,通过率均约98%。
  • AC任务ME为-0.026±0.06 Gy、SSIM为99.5±0.25%、PSNR为47.28±3.31 dB。
  • SC任务ME为-0.014±0.05 Gy、SSIM为99.88±0.099%、PSNR为55.9±4 dB。
  • ASC任务ME为-0.04±0.06 Gy、SSIM为99.57±0.33%、PSNR为47.97±3.6 dB。
  • 肿瘤MAE在AC、SC、ASC任务分别为7.22±5.9 Gy、8±9.3 Gy、11.8±12.02 Gy;全正常肝脏MAE分别为1.09±0.86 Gy、0.9±0.8 Gy、1.3±0.98 Gy。

临床怎么用

  • 作者认为,这些任务特异性模型在CT图像不可用或因配准不良不可靠时,经更大数据集训练后可能有助于完成必要的校正。

局限性

  • 作者指出模型需在更大数据集训练后才能应用。

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)。

原文

BACKGROUND: This work aimed to develop deep learning (DL) models for CT-free attenuation and Monte Carlo-based scatter correction (AC, SC) in quantitative 90Y SPECT imaging for improved dose calculation. METHODS: Data of 190 patients who underwent 90Y selective internal radiation therapy (SIRT) with glass microspheres was studied. Voxel-level dosimetry was performed on uncorrected and corrected SPECT images using the local energy deposition method. Three deep learning models were trained individually for AC, SC, and joint ASC using a modified 3D shifted-window UNet Transformer (Swin UNETR) architecture. Corrected and unorrected dose maps served as reference and as inputs, respectively. The data was split into train set (~ 80%) and unseen test set (~ 20%). Training was conducted in a five-fold cross-validation scheme. The trained models were tested on the unseen test set. The model's performance was thoroughly evaluated by comparing organ- and voxel-level dosimetry results between the reference and DL-generated dose maps on the unseen test dataset. The voxel and organ-level evaluations also included Gamma analysis with three different distances to agreement (DTA (mm)) and dose difference (DD (%)) criteria to explore suitable criteria in SIRT dosimetry using SPECT. RESULTS: The average ± SD of the voxel-level quantitative metrics for AC task, are mean error (ME (Gy)): -0.026 ± 0.06, structural similarity index (SSIM (%)): 99.5 ± 0.25, and peak signal to noise ratio (PSNR (dB)): 47.28 ± 3.31. These values for SC task are - 0.014 ± 0.05, 99.88 ± 0.099, 55.9 ± 4, respectively. For ASC task, these values are as follows: -0.04 ± 0.06, 99.57 ± 0.33, 47.97 ± 3.6, respectively. The results of voxel level gamma evaluations with three different criteria, namely "DTA: 4.79, DD: 1%", "DTA:10 mm, DD: 5%", and "DTA: 15 mm, DD:10%" were around 98%. The mean absolute error (MAE (Gy)) for tumor and whole normal liver across tasks are as follows: 7.22 ± 5.9 and 1.09 ± 0.86 for AC, 8 ± 9.3 and 0.9 ± 0.8 for SC, and 11.8 ± 12.02 and 1.3 ± 0.98 for ASC, respectively. CONCLUSION: We developed multiple models for three different clinically scenarios, namely AC, SC, and ASC using the patient-specific Monte Carlo scatter corrected and CT-based attenuation corrected images. These task-specific models could be beneficial to perform the essential corrections where the CT images are either not available or not reliable due to misalignment, after training with a larger dataset.

来源:Eur J Nucl Med Mol Imaging,查看原文。版权归原作者所有。

引用

  1. Mansouri Z, Salimi Y, Wolf NB, et al. CT-free attenuation and Monte-Carlo based scatter correction-guided quantitative 90Y-SPECT imaging for improved dose calculation using deep learning. Eur J Nucl Med Mol Imaging. 2025. PMID: 40080141. doi:10.1007/s00259-025-07191-5

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。