The Value of Cardiovascular Imaging Methods in the Diagnosis of Complex Congenital Heart Disease Evaluated by ROC Analysis

  • Guo-Zhen Chen, Department of Pediatric Cardiology, Shanghai Children’s Medical Center, Medical College of Shanghai Jiaotong University, China
  • Dr Kun Sun, Department of Pediatric Cardiology, Shanghai Children’s Medical Center, China
  • Yu-Min Zhong, Department of Pediatric Cardiology, Shanghai Children’s Medical Center, Medical College of Shanghai Jiaotong University, China
  • Lan-Ping Wu, Department of Pediatric Cardiology, Shanghai Children’s Medical Center, Medical College of Shanghai Jiaotong University, China
  • Ai-Min Sun, Department of Pediatric Cardiology, Shanghai Children’s Medical Center, Medical College of Shanghai Jiaotong University, China
  • Yu-Qi Zhang, Department of Pediatric Cardiology, Shanghai Children’s Medical Center, Medical College of Shanghai Jiaotong University, China
  • Qian Wang, Department of Pediatric Cardiology, Shanghai Children’s Medical Center, Medical College of Shanghai Jiaotong University, China
  • Min Zhu, Department of Pediatric Cardiology, Shanghai Children’s Medical Center, Medical College of Shanghai Jiaotong University, China
  • Objective To evaluate the diagnostic value of imaging methods in pathological morphology of complex congenital heart disease (CHD) using receiver operating characteristic (ROC) analysis. Methods Twenty-eight children (aged 33.7±48.1 months) with CHD were diagnosed by three clinical-data-blinded cardiologists with two-dimensional echocardiography (2DE), real-time three-dimensional echocardiography (RT-3DE), magnetic resonance imaging (MRI) and digital subtraction angiography (DSA), respectively. With surgical findings as “Golden Standard”, the results of cardiovascular malformations were compared and grouped by a five-point categorical scale ranged from definitely abnormal to normal and further analyzed by ROC curve. Results To diagnose the intracardiac malformations in atria, ventricles, great arteries, atrio-ventricular and ventricular-arterial continuity, ROC curve for RT-3DE was located left-superior to that for 2DE, MRI and DSA. The area under ROC curve for RT-3DE (0.96) was higher than that for 2DE (0.92), MRI (0.91) and DSA (0.89). There was significant difference by the method of Z test between RT-3DE and 2DE (Z=2.30, P=0.0215), MRI (Z=2.80, P=0.0050), DSA (Z=3.18, P=0.0015). While for the diagnosis of extracardiac malformations, ROC curve for DSA was located left-superior to that for MRI and 2DE. The area under ROC curve for DSA (0.99) was higher than that for MRI (0.98) and 2DE (0.92). There was significant difference between DSA and 2DE (Z=2.67, P=0.0075), but no significant difference between DSA and MRI (Z=1.52, P=0.1292). Conclusions: ROC analysis showed that RT-3DE had advantages in displaying spatial intracardiac structures, while DSA and MRI had advantages in displaying extracardiac structures, MRI could replace angiography to some extent.