Lopes Leonor, Jiao Fangyang, Xue Song, Pyka Thomas, Krieger Korbinian, Ge Jingjie, Xu Qian, Fahmi Rachid, Spottiswoode Bruce, Soliman Ahmed, Buchert Ralph, Brendel Matthias, Hong Jimin, Guan Yihui, Bassetti Claudio L A, Rominger Axel, Zuo Chuantao, Shi Kuangyu, Wu Ping
Department of Nuclear Medicine, Inselspital, Bern University Hospital, University of Bern, Freiburgstrasse 18, Bern, 3010, Switzerland.
Graduate School for Cellular and Biomedical Sciences, University of Bern, Bern, Switzerland.
Eur J Nucl Med Mol Imaging. 2025 Feb;52(3):851-863. doi: 10.1007/s00259-024-06961-x. Epub 2024 Nov 19.
Dopamine transporter imaging is routinely used in Parkinson's disease (PD) and atypical parkinsonian syndromes (APS) diagnosis. While [C]CFT PET is prevalent in Asia with a large APS database, Europe relies on [I]FP-CIT SPECT with limited APS data. Our aim was to develop a deep learning-based method to convert [C]CFT PET images to [I]FP-CIT SPECT images, facilitating multicenter studies and overcoming data scarcity to promote Artificial Intelligence (AI) advancements.
A CycleGAN was trained on [C]CFT PET (n = 602, 72%PD) and [I]FP-CIT SPECT (n = 1152, 85%PD) images from PD and non-parkinsonian control (NC) subjects. The model generated synthetic SPECT images from a real PET test set (n = 67, 75%PD). Synthetic images were quantitatively and visually evaluated.
Fréchet Inception Distance indicated higher similarity between synthetic and real SPECT than between synthetic SPECT and real PET. A deep learning classification model trained on synthetic SPECT achieved sensitivity of 97.2% and specificity of 90.0% on real SPECT images. Striatal specific binding ratios of synthetic SPECT were not significantly different from real SPECT. The striatal left-right differences and putamen binding ratio were significantly different only in the PD cohort. Real PET and real SPECT had higher contrast-to-noise ratio compared to synthetic SPECT. Visual grading analysis scores showed no significant differences between real and synthetic SPECT, although reduced diagnostic performance on synthetic images was observed.
CycleGAN generated synthetic SPECT images visually indistinguishable from real ones and retained disease-specific information, demonstrating the feasibility of translating [C]CFT PET to [I]FP-CIT SPECT. This cross-modality synthesis could enhance further AI classification accuracy, supporting the diagnosis of PD and APS.
多巴胺转运体成像常用于帕金森病（PD）和非典型帕金森综合征（APS）的诊断。虽然[C]CFT正电子发射断层扫描（PET）在亚洲较为普遍且有大量APS数据库，但欧洲依赖[I]FP-CIT单光子发射计算机断层扫描（SPECT）且APS数据有限。我们的目标是开发一种基于深度学习的方法，将[C]CFT PET图像转换为[I]FP-CIT SPECT图像，以促进多中心研究并克服数据稀缺问题，推动人工智能（AI）发展。
使用来自PD和非帕金森病对照（NC）受试者的[C]CFT PET（n = 602，72%为PD）和[I]FP-CIT SPECT（n = 1152，85%为PD）图像训练循环一致对抗网络（CycleGAN）。该模型从真实PET测试集（n = 67，75%为PD）生成合成SPECT图像。对合成图像进行定量和视觉评估。
弗雷歇因ception距离表明，合成SPECT与真实SPECT之间的相似度高于合成SPECT与真实PET之间的相似度。在合成SPECT上训练的深度学习分类模型在真实SPECT图像上的灵敏度为97.2%，特异性为90.0%。合成SPECT的纹状体特异性结合率与真实SPECT无显著差异。仅在PD队列中，纹状体左右差异和壳核结合率有显著差异。与合成SPECT相比，真实PET和真实SPECT具有更高的对比噪声比。视觉分级分析评分显示真实SPECT与合成SPECT之间无显著差异，尽管在合成图像上观察到诊断性能有所下降。
CycleGAN生成的合成SPECT图像在视觉上与真实图像难以区分，并保留了疾病特异性信息，证明了将[C]CFT PET转换为[I]FP-CIT SPECT的可行性。这种跨模态合成可以提高AI分类的准确性，支持PD和APS的诊断。
