文献速递心血管CT

深度学习模型用于低剂量CT晚期碘增强成像及细胞外容积定量

Deep learning model for low-dose CT late iodine enhancement imaging and extracellular volume quantification.
PubMed 文献 · Eur Radiol · 2024-12-20

本研究回顾性纳入两家医院因胸部不适行CT心肌灌注+CT血管成像+晚期碘增强检查的患者423例,开发并验证残差密集网络(RDN)和条件生成对抗网络(cGAN)两种深度学习去噪模型。RDN模型生成的去噪晚期碘增强图像信噪比和对比噪声比均显著高于cGAN模型和原始图像,并显著提高视觉分析可识别性,且基于去噪单层图像可实现准确的CT细胞外容积定量。

要点

  • RDN去噪图像信噪比20.5±4.7、对比噪声比7.5±2.3,均高于cGAN图像和原始图像。
  • 节段水平上RDN图像评估晚期碘增强的AUC显著高于cGAN和原始图像。
  • 基于RDN去噪单层图像的细胞外容积定量AUC和准确率显著高于cGAN和原始图像。

核心结论

  • RDN模型生成的去噪晚期碘增强图像信噪比和对比噪声比显著高于cGAN模型和原始图像,显著提高视觉分析可识别性,且使用去噪单层图像可实现准确的CT细胞外容积定量。

研究设计与样本

  • 回顾性研究,纳入两家医院因胸部不适行CT心肌灌注+CT血管成像+晚期碘增强检查的患者423例,随机分为训练组182例、调优组48例、内部验证组92例和外部验证组101例,开发并验证RDN和cGAN两种深度学习模型。

诊断标准与关键数值

  • 以细胞外容积>30%为增高;RDN图像信噪比20.5±4.7、对比噪声比7.5±2.3,cGAN图像信噪比13.3±1.9、对比噪声比4.5±1.1,原始图像信噪比4.4±0.7、对比噪声比1.6±0.4;节段水平RDN图像评估晚期碘增强的AUC显著高于cGAN图像(p=0.040)和原始图像(p<0.001);RDN细胞外容积的AUC和准确率显著高于cGAN和原始图像(p均<0.001)。

局限性

  • 原文未报告局限性。

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

OBJECTIVES: To develop and validate deep learning (DL)-models that denoise late iodine enhancement (LIE) images and enable accurate extracellular volume (ECV) quantification.

METHODS: This study retrospectively included patients with chest discomfort who underwent CT myocardial perfusion + CT angiography + LIE from two hospitals. Two DL models, residual dense network (RDN) and conditional generative adversarial network (cGAN), were developed and validated. 423 patients were randomly divided into training (182 patients), tuning (48 patients), internal validation (92 patients) and external validation group (101 patients). LIEsingle (single-stack image), LIEaveraging (averaging multiple-stack images), LIERDN (single-stack image denoised by RDN) and LIEGAN (single-stack image denoised by cGAN) were generated. We compared image quality score, signal-to-noise (SNR) and contrast-to-noise (CNR) of four LIE sets. The identifiability of denoised images for positive LIE and increased ECV (> 30%) was assessed.

RESULTS: The image quality of LIEGAN (SNR: 13.3 ± 1.9; CNR: 4.5 ± 1.1) and LIERDN (SNR: 20.5 ± 4.7; CNR: 7.5 ± 2.3) images was markedly better than that of LIEsingle (SNR: 4.4 ± 0.7; CNR: 1.6 ± 0.4). At per-segment level, the area under the curve (AUC) of LIERDN images for LIE evaluation was significantly improved compared with those of LIEGAN and LIEsingle images (p = 0.040 and p < 0.001, respectively). Meanwhile, the AUC and accuracy of ECVRDN were significantly higher than those of ECVGAN and ECVsingle at per-segment level (p < 0.001 for all).

CONCLUSIONS: RDN model generated denoised LIE images with markedly higher SNR and CNR than the cGAN-model and original images, which significantly improved the identifiability of visual analysis. Moreover, using denoised single-stack images led to accurate CT-ECV quantification.

KEY POINTS: Question Can the developed models denoise CT-derived late iodine enhancement high images and improve signal-to-noise ratio? Findings The residual dense network model significantly improved the image quality for late iodine enhancement and enabled accurate CT- extracellular volume quantification. Clinical relevance The residual dense network model generates denoised late iodine enhancement images with the highest signal-to-noise ratio and enables accurate quantification of extracellular volume.

来源:Eur Radiol,查看原文。版权归原作者所有。

引用

  1. Yu Y, Wu D, Lan Z, et al. Deep learning model for low-dose CT late iodine enhancement imaging and extracellular volume quantification. Eur Radiol. 2024. PMID: 39704803. doi:10.1007/s00330-024-11288-0

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。