文献速递妇产影像神经MR

深度学习超分辨率重建在胎儿脑T2加权SSFSE成像中的评估

Evaluation of deep learning-based super-resolution reconstruction in fetal brain T2-weighted single-shot fast spin-echo imaging.
PubMed 文献 · Neuroradiology · 2026-10-29

该研究纳入14例胎儿脑MRI的SSFSE序列,比较有无深度学习PIQE重建(3倍放大)的图像质量。PIQE显著提高了SNR、CNR和FWHM锐度指标(p<0.001或p<0.01),两名放射科医师的主观评分(噪声、对比、锐度、伪影、整体质量)也均显著更优(p<0.001),结论认为PIQE有潜力提升胎儿脑SSFSE图像质量。

要点

  • PIQE深度学习重建可显著提升胎儿脑SSFSE图像的SNR、CNR和锐度。
  • 主观评分显示PIQE在噪声、对比、锐度、伪影和整体质量上均更优。
  • 观察者间一致性中等至优秀(κ=0.52~0.82),锐度和整体质量一致性最高。

关键影像表现/征象

  • PIQE重建图像的SNR、CNR和FWHM均显著优于无PIQE图像(SNR、CNR:p<0.001;FWHM:p<0.01)。
  • 两名放射科医师对PIQE图像的噪声、对比、锐度、伪影和整体质量主观评分均显著更高(p<0.001)。

诊断标准与关键数值

  • 主观评分采用4分制,评估噪声、对比、锐度、伪影和整体图像质量。
  • 观察者间一致性Kappa值:噪声κ=0.52,对比κ=0.74,锐度κ=0.82,伪影κ=0.61,整体质量κ=0.80。

临床应用与阅片要点

  • 在胎儿脑SSFSE成像中,可考虑应用PIQE深度学习重建以改善图像锐度和整体质量。
  • 主观评估时需注意噪声和伪影方面的一致性仅为中等(κ=0.52和0.61)。

局限性

  • 样本量较小(仅14例),可能影响结论的普适性。
  • 研究为单中心回顾性设计,未报告外部验证结果。

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

PURPOSE: The study aimed to evaluate the efficacy of a deep learning-based Precise IQ Engine (PIQE) in enhancing image sharpness and quality during MRI reconstruction to enhance the image quality of fetal brain single-shot fast spin-echo (SSFSE) imaging.

METHODS: Fetal brain MRI scans of 14 patients, including an SSFSE sequence, acquired between January 2024 and February 2025, were analyzed. Images were reconstructed with and without PIQE (zoom ratio of 3 from low-resolution matrices). We quantified the signal-to-noise ratio (SNR), contrast-to-noise ratio (CNR), and image sharpness using the full width at half maximum (FWHM) of the fetal brain. Two board-certified radiologists assessed the image noise, contrast, sharpness, artifacts, and overall image quality of the two image types on a 4-point scale. We compared the quantitative and qualitative scores between the images with and without PIQE using the Wilcoxon signed-rank test.

RESULTS: SNRs, CNRs, and FWHM were significantly enhanced with PIQE compared to those without PIQE (SNR, CNR: p < 0.001, FWHM : p < 0.01). Qualitative scores for noise (p < 0.001), contrast (p < 0.001), sharpness (p < 0.001), artifacts (p < 0.001), and overall image quality (p < 0.001) were superior in images obtained with PIQE compared to those without it. Kappa analysis revealed moderate-to- excellent agreement for different image quality aspects: noise (κ = 0.52), contrast (κ = 0.74), sharpness (κ = 0.82), artifacts (κ = 0.61), and overall image quality (κ = 0.80).

CONCLUSION: PIQE potentially enhances the image quality of fetal brain MRI using an SSFSE sequence.

来源:Neuroradiology,查看原文。版权归原作者所有。

引用

  1. Sato Y, Mori H, Yamazaki H, et al. Evaluation of deep learning-based super-resolution reconstruction in fetal brain T2-weighted single-shot fast spin-echo imaging. Neuroradiology. 2026. PMID: 42955312. doi:10.1007/s00234-026-04218-y

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。