心脏磁共振自由呼吸运动校正人工智能电影序列在心力衰竭患者中的临床应用

Clinical application of free breathing with motion correction artificial intelligence cine sequence of cardiac magnetic resonance in patients with heart failure

目的通过评估自由呼吸运动校正人工智能(free-breathing with motion correction artificial intelligence,FB-MOCO AI)电影序列的图像质量、双心室容积和功能参数,探讨其在心力衰竭患者心脏磁共振(cardiac magnetic resonance,CMR)检查中应用的可行性。材料与方法前瞻性连续纳入2022年8月至2023年5月期间来我院进行CMR检查的心力衰竭患者29例。所有患者均行常规屏气电影和FB-MOCO AI电影序列的心脏短轴面扫描。由两名放射科医师对两种电影序列采集的图像进行整体图像质量评分(5分法)。测量两种序列的双心室容积和功能参数,包括舒张末期容积(end-diastolic volume,EDV)、收缩末期容积(end-systolic volume,ESV)、每搏输出量(stroke volume,SV)、射血分数(ejection fraction,EF)和左心室舒张末期质量(LV end-diastolic mass,LVEDM)。采用配对样本t检验或Wilcoxon符号秩检验来评估两种序列之间定量数据和定性评分数据的差异。并对两种序列间定量参数做相关性分析。通过Bland-Altman分析来评估从不同序列中获得的定量参数的一致性。结果FB-MOCO AI电影序列的总扫描时间[(14.3±1.9)s]比常规屏气电影[(79.2±11.4)s]明显缩短(P<0.001)。FB-MOCO AI电影[(4.4±0.5)分]的总体图像质量评分与常规屏气电影[(4.3±0.7)分]相当,差异无统计学意义(P=0.343)。两种电影序列间所有测量的双心室容积和功能参数以及LVEDM差异均无统计学意义(P值均>0.05),并且有强相关性(r值均≥0.94,P值均<0.001)。Bland-Altman分析显示,两种电影序列的所有心功能定量参数平均差异均接近于零,且变异范围很小,一致性高。结论与常规屏气电影序列相比,心衰患者中FB-MOCO AI电影序列扫描时间明显缩短并且提供了相当的图像质量、心室容积和功能指标,有潜力替代心衰患者的常规电影序列应用于临床。

Objective:To explore the feasibility of free-breathing with motion correction artificial intelligence(FB-MOCO AI)cine sequence of cardiac magnetic resonance(CMR)in patients with heart failure by evaluating the image quality,biventricular volumetric and functional parameters.Materials and Methods:From August 2022 to May 2023,29 patients with heart failure who underwent CMR in our hospital were prospectively consecutive enrolled.All patients underwent cardiac short-axis scans with conventional breath-hold(BH)cine and FB-MOCO AI cine sequences.The overall image quality of the two cine sequences was scored by two radiologists(5-point scale).The parameters of biventricular volume and function including end-diastolic volume(EDV),end-systolic volume(ESV),stroke volume(SV),ejection fraction(EF),and left ventricular end-diastolic mass(LVEDM)were measured.Paired t-test or Wilcoxon signed-rank test were used to assess differences in quantitative data and qualitative scoring data between the two sequences.The correlation analysis of the quantitative parameters between the two sequences was also assessed.Bland-Altman analysis was performed to assess the agreement of quantitative parameters obtained from different sequences.Results:The durations of FB-MOCO AI cine[(14.3±1.9)s]scans were significantly shorter than conventional BH cine[(79.2±11.4)s](P˂0.001).Overall image quality of FB-MOCO AI cine(4.4±0.5)were comparable to conventional BH cine(4.3±0.7)(P=0.343).All measured biventricular volumetric and functional parameters and LVEDM of two cine sequences presented no statistical difference(P˃0.05 for all),and there was excellent correlations(r˃0.94,P˂0.001 for all).Bland-Altman analysis showed that all the quantitative parameters between the two cine sequences had excellent agreement with a mean difference close to zero and a small range of variation.Conclusions:Compared to conventional breath-hold cine,the FB-MOCO AI cine method in patients with heart failure achieved comparable image quality and biventricular volumetric and functional parameters with shortened scan times,suggesting it has the potential to replace conventional cine sequence in patients with heart failure for clinical application.