Skip to main content

Table 7 Performance comparison of the proposed segmentation on ISBI 2016 and 2017 and PH2 dataset when using Jaccard coefficient or cross-entropy loss for training

From: DermoNet: densely linked convolutional neural network for efficient skin lesion segmentation

Dataset

Method

JC (%)

ISBI 2016

D e r m o N e t ( C E )

79.16

 

D e r m o N e t ( J C )

82.5

PH2

D e r m o N e t ( C E )

72.14

 

D e r m o N e t ( J C )

85.3

ISBI 2017

D e r m o N e t ( C E )

75.4

 

D e r m o N e t ( J C )

78.3

  1. Here, JC denotes Jaccard coefficient and CE denotes cross-entropy, respectively