Table 4 Comparison of segmentation performance between deep learning networks for combined dataset of three major vessels. The highest F1 score was shown in bold.

Re

Pr

F1

Re

Pr

F1

Re

Pr

F1

Re

Pr

F1

Total (n = 3302)

0.869 ± 0.151

0.885 ± 0.125

0.871 ± 0.130

0.915 ± 0.117^*

0.916 ± 0.106^*

0.913 ± 0.108^*

0.921 ± 0.112^*†§

0.918 ± 0.103^*

0.917 ± 0.103^*†

0.915 ± 0.116^*

0.920 ± 0.105^*†‡

0.915 ± 0.107^*

RCA (n = 1021)

0.915 ± 0.112

0.929 ± 0.076

0.918 ± 0.086

0.941 ± 0.084^*

0.939 ± 0.068^*

0.937 ± 0.071^*

0.945 ± 0.071^*§

0.940 ± 0.064^*

0.940 ± 0.058^*

0.943 ± 0.072^*

0.942 ± 0.063^*

0.940 ± 0.058^*

LAD (n = 1439)

0.872 ± 0.141

0.882 ± 0.114

0.872 ± 0.119

0.921 ± 0.104^*

0.917 ± 0.090^*

0.916 ± 0.094^*

0.928 ± 0.094^*†§

0.920 ± 0.085^*

0.922 ± 0.084^*

0.919 ± 0.105^*

0.922 ± 0.094^*†‡

0.918 ± 0.095^*

LCX (n = 842)

0.809 ± 0.186

0.837 ± 0.166

0.813 ± 0.166

0.875 ± 0.156^*

0.887 ± 0.151^*

0.878 ± 0.150^*

0.878 ± 0.158^*

0.888 ± 0.150^*

0.879 ± 0.153^*

0.875 ± 0.158^*

0.890 ± 0.148^*

0.879 ± 0.151^*

Re

Pr

F1

Re

Pr

F1

Re

Pr

F1

Re

Pr

F1

Total (n = 181)

0.764 ± 0.251

0.855 ± 0.202

0.791 ± 0.226

0.846 ± 0.232^*

0.871 ± 0.191

0.849 ± 0.214^*

0.898 ± 0.155^*

0.904 ± 0.126^*

0.896 ± 0.138^*

0.887 ± 0.172^*

0.905 ± 0.144^*†

0.890 ± 0.161^*†

RCA (n = 63)

0.876 ± 0.172

0.934 ± 0.051

0.891 ± 0.126

0.918 ± 0.135

0.921 ± 0.075

0.911 ± 0.112

0.931 ± 0.096^*

0.925 ± 0.062

0.924 ± 0.066

0.939 ± 0.072^*

0.928 ± 0.073

0.930 ± 0.058^*

LAD (n = 66)

0.788 ± 0.205

0.855 ± 0.173

0.814 ± 0.183

0.882 ± 0.185^*

0.876 ± 0.179

0.878 ± 0.179^*

0.914 ± 0.127^*

0.901 ± 0.126

0.906 ± 0.122^*

0.887 ± 0.146^*

0.915 ± 0.090^*†‡

0.892 ± 0.140^*

LCX (n = 52)

0.599 ± 0.299

0.758 ± 0.292

0.638 ± 0.284

0.714 ± 0.314^*

0.805 ± 0.271

0.738 ± 0.294^*

0.836 ± 0.218^*†

0.884 ± 0.174^*

0.848 ± 0.197^*†

0.825 ± 0.252^*†

0.864 ± 0.232^*

0.837 ± 0.239^*†