Table 2 Image segmentation performance compared with SOTA models

From: Large language models driven neural architecture search for universal and lightweight disease diagnosis on histopathology slide images

Search strategy

BCSS

PanNuke

 

Dice (%)↑

IoU (%)↑

FLOPs (G)↓

Params (M)↓

Dice (%)↑

IoU (%)↑

FLOPs (G)↓

Params (M)↓

U-Net

71.56

56.33

14.80

18.44

84.93

74.99

14.80

18.44

FPN

72.30

57.21

17.00

11.49

88.45

80.07

17.00

11.49

Pathology-NAS U-Net(ours)

74.33

59.68

10.58

11.37

89.24

81.25

14.33

8.34

  1. For U-Net backbone, Pathology-NAS is compared with U-Net and FPN in terms of dice score, IoU score, FLOPs and Params. The best performance is highlighted in bold. Among all methods, Pathology-NAS holds the optimal segmentation performance with the lowest FLOPs and parameters.
Back to article page