Skip to main content

Table 2 Performance of the discriminations for spectral fitting (SF) [7], intra-class correlation (ICC) [9, 10], and the spatio-spectral mean-shift (SSM-S) algorithm, considering two evaluation criteria: the percentage of well-clustered pixels (where WCP is the value) and the Dice similarity coefficient (DSC)

From: Spatial and spectral regularization to discriminate tissues using multispectral photoacoustic imaging

Dataset

Desired discrimination

Normalization

WCP (%)

DSC

   

SF

ICC

SSM-S

SF

ICC

SSM-S

Liver

Fresh/stale

No

95

76

100

0.91

0.63

0.87

Blood and ink

None (concentrations)

Yes

86

87

98

0.86

0.87

0.98

Blood and ink

Concentrations

No

77

34

99

0.77

0.34

0.97