Table 2 Quantitative comparison with benchmark methods.

From: Artificial intelligence-assisted fast screening cervical high grade squamous intraepithelial lesion and squamous cell carcinoma diagnosis and treatment planning

 

Target

Data type

Aver. size (pixel)

Precision

Recall

F-measure

Jaccard

Proposed method

HSIL

WSIs

91257 41546

0.93

0.90

0.88

0.84

U-net19

HSIL

WSIs

91257 41546

0.15

0.70

0.17

0.12

SegNet20

HSIL

WSIs

91257 41546

0.26

0.88

0.27

0.23

AraĂşjo et al.17

LSIL/ASCUS

Image patches

1383 1036

0.66

0.72

0.69

0.53

  1. The proposed method is significantly better than the benchmark approaches (p<0.0001).
Back to article page