Table 3 Results of ablation tests for the proposed model in predicting 4-category molecular subtypes of breast cancer in the test cohort (n = 672).

From: Predicting breast cancer types on and beyond molecular level in a multi-modal fashion

Method

Modality

Accuracy (%)

Precision (%)

Recall (%)

F1-score

MCC

Multi-ResNet50

MG + US

84.4 [81.5, 87.1]

83.7 [80.4, 86.9]

81.3 [77.8, 84.5]

0.820 [0.786, 0.852]

0.777 [0.736, 0.814]

MulR-interSA

MG + US

85.4 [82.6, 88.1]

84.7 [81.5, 87.8]

82.3 [79.1, 85.6]

0.830 [0.797, 0.862]

0.793 [0.753, 0.829]

MulR-iiSA

MG + US

86.1 [83.6, 88.7]

86.1 [83.1, 88.9]

83.0 [79.8, 86.2]

0.839 [0.805, 0.870]

0.803 [0.767, 0.839]

MulR-interCSA

MG + US

87.5 [85.0, 90.0]

87.2 [84.3, 89.9]

84.6 [81.5, 87.7]

0.853 [0.823, 0.884]

0.822 [0.786, 0.858]

Proposed (MDL-IIA)

MG + US

88.5 [86.0, 90.9]

87.8 [85.0, 90.7]

85.4 [82.2, 88.4]

0.862 [0.831, 0.892]

0.837 [0.803, 0.870]

  1. Values in brackets are 95% confidence intervals [95% CI, %].
  2. MG mammography, US ultrasound, MulR Multi-ResNet, SA self-attention, iiSA intra- and inter-self-attention, CSA channel and spatial attention, interSA inter self-attention, interCSA inter channel and spatial attention, MCC matthews correlation coefficient, MDL-IIA multi-modal deep learning with intra- and inter-modality attention modules.
Back to article page