Skip to main content

Table 3 Automatic segmentation performance in terms of DSC and sensitivity

From: Automatic segmentation and classification of breast lesions through identification of informative multiparametric PET/MRI features

Features DSC mean ± SD/median ± IQR Sensitivity mean ± SD/median ± IQR Number of features
GI 0.607 ± 0.238/0.691 ± 0.255 0.661 ±  0.234/0.721 ± 0.263 2
GI without PET 0.608 ± 0.281/0.722 ± 0.419 0.739 ±  0.290/0.876 ± 0.238 11
GI without DWI 0.573 ± 0.290/0.708 ± 0.445 0.669 ±  0.324/0.825 ± 0.445 8
GI without DWI, without PET 0.577 ± 0.277/0.665 ± 0.419 0.658 ±  0.343/0.813 ± 0.597 25
mRMR 0.665 ± 0.236/0.757 ± 0.189 0.743 ±  0.267/0.836 ± 0.269 8
mRMR without PET 0.618 ± 0.272/0.749 ± 0.413 0.748 ±  0.277/0.873 ± 0.307 7
mRMR without DWI 0.601 ± 0.268/0.701 ± 0.377 0.686 ±  0.328/0.801 ± 0.538 7
mRMR without DWI, without PET 0.584 ± 0.300/0.710 ± 0.393 0.613 ±  0.338/0.784 ± 0.496 8
  1. DSC Dice similarity coefficient, DWI diffusion-weighted imaging, IQR interquartile range, GI Gini Importance, mRMR minimum-redundancy-maximum-relevance, PET positron emission tomography, SD standard deviation. Values presented in bold are the highest values