文献速递神经CTMR

CT风格迁移合成FLAIR提升脑转移瘤周水肿显示

CT-to-FLAIR Style Transfer Improves Perilesional Edema Conspicuity in Brain Metastases.
PubMed 文献 · AJNR Am J Neuroradiol · 2026-06-11

回顾性分析321例有或无脑转移患者,利用生成对抗网络将平扫CT转换为类T2 FLAIR图像。比较三种UNet变体后,UNet++合成质量最好;12名神经放射科医生评估显示,合成MR较CT显著提高病灶显示,较真实MR略差但图像质量无显著差异。

要点

  • GAN可将平扫CT转为类T2 FLAIR图像,提升脑转移瘤周水肿的显示能力。
  • 合成FLAIR病灶显示评分高于CT,但真实性偏好仍以真实MR为主(73.3%)。
  • UNet++生成的合成MR图像质量与真实MR无显著差异,可供阅片时辅助观察。

核心结论

  • 基于GAN的CT风格迁移可合成类T2 FLAIR图像,较平扫CT显著提高脑转移灶水肿显示度。
  • 合成MR图像质量与真实MR无显著差异,但读者认为真实MR更具真实性。

研究设计与样本

  • 回顾性研究,共321例患者(235例有脑转移,86例无),配对平扫CT与T2 FLAIR MR图像。
  • 采用5折交叉验证训练GAN,判别器为PatchGAN,比较Attention-UNet、SUNet、UNet++三种生成器。
  • 选取最佳生成器生成的20幅合成MR图像,由12名神经放射科医生以5分制评分。

关键影像表现/征象

  • UNet++合成图像质量最佳,SSIM=0.9119±0.0008,MAE=0.1232±0.0022,MSE=0.0201±0.0004。
  • 合成MR病灶显示评分4.34±0.18星,高于CT的3.16±0.44星,低于真实MR的4.67±0.13星。
  • 与CT相比,合成MR提高病灶显示的配对差值为1.18±0.39,效应量d=2.2。

诊断标准与关键数值

  • 合成MR较CT显著提高病灶显示(p<0.001),平均配对差值1.18±0.39,Cohen's d=2.2。
  • 合成MR较真实MR病灶显示略降低(p<0.001),平均配对差值0.33±0.23,Cohen's d=1.6。
  • 读者在73.3%的比较中选择真实MR更真实(p<0.001),但合成MR图像质量未显著劣于真实MR(p=0.078)。

临床应用与阅片要点

  • 平扫CT经风格迁移生成的合成FLAIR可用于提高脑转移瘤周水肿的检出与显示。
  • 对不能接受MR检查的患者,该技术可提供接近FLAIR对比的图像信息辅助诊断。
  • 阅片时需注意合成图像可能降低真实性感知,仍应以真实MR作为首选参考。

局限性

  • 回顾性单中心研究,样本来源和评估方式存在选择偏倚可能。
  • 仅评估平扫CT与T2 FLAIR配对,未涉及增强后病灶活性或对比增强信息。

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

BACKGROUND AND PURPOSE: To evaluate whether style transfer using generative adversarial networks (GANs) can synthesize T2 FLAIR-like MR images from non-contrast CT to enhance lesion conspicuity of brain metastases.

MATERIALS AND METHODS: This retrospective study analyzed 321 patients (235 with brain metastases, 86 without) whose non-contrast CT head and T2 FLAIR MR images were paired based on lesion stability and temporal proximity. Following pre-processing, CT-MR pairs were used to train GANs for CT-to-synthetic-FLAIR style transfer with 5-fold cross-validation. PatchGAN discriminator was used. 3 UNet-family generators (Attention-UNet, SUNet, UNet++) were compared using mean-absolute-error (MAE), mean-squared-error (MSE), and structural similarity index measure (SSIM). 20 synthetic MR images generated by the best performing generator were shown to 12 neuroradiologists, who rated image quality and lesion conspicuity for CT, real MR, and synthetic MR (scale 1-5★), and realism preference between real and synthetic MR.

RESULTS: UNet++ produced the best image quality, with SSIM = 0.9119 ± 0.0008, MAE = 0.1232 ± 0.0022, and MSE = 0.0201 ± 0.0004 (95% CIs, all p<0.001). Mean reader-aggregated image quality ratings were 4.28 ± 0.15★ for real MR and 4.10 ± 0.14★ for synthetic MR; mean lesion conspicuity ratings were 4.67 ± 0.13★ for real MR, 4.34 ± 0.18★ for synthetic MR, and 3.16 ± 0.44★ for CT (all 95% CIs). Synthetic MR significantly improved lesion conspicuity over CT (p<0.001), with mean paired difference (MPD) 1.18 ± 0.39 and Cohen's d=2.2; while synthetic MR showed slightly reduced conspicuity compared to real MR (p<0.001), with MPD 0.33 ± 0.23 and Cohen's d=1.6 (95% CIs). Readers selected real MR as more realistic than synthetic MR in 73.3% of comparisons (p<0.001), but synthetic image quality was not significantly inferior to real MR (p=0.078).

CONCLUSION: GAN-generated synthetic T2 FLAIR images from CT improve edema conspicuity of brain metastases relative to CT.

来源:AJNR Am J Neuroradiol,查看原文。版权归原作者所有。

引用

  1. Hsu, Narayanan, McRae-Posani, et al. CT-to-FLAIR Style Transfer Improves Perilesional Edema Conspicuity in Brain Metastases. AJNR Am J Neuroradiol. 2026. PMID: 42242726. doi:10.3174/ajnr.A9457

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。