Table 2 Classification performance of the FFNN algorithm. Here, n is the sample size of the test set, Th is the value of the classification threshold, SE denotes the sensitivity, SP is the specificity, and AUROC is the area under the receiver operating characteristic curve.

From: Hyperspectral image processing for the identification and quantification of lentiviral particles in fluid samples

 

FFNN

WET

DRY Residue

n

Th

SE (%)

SP (%)

AUROC

n

Th

SE (%)

SP (%)

AUROC

PBS

Pixel

15,104

0.90

75.6

76.0

0.84

20,527

0.91

54.0

52.7

0.54

Droplet

13

0.52

100.0

100.0

1.00

13

0.16

87.5

80.0

0.88

AS

Pixel

28,646

0.84

66.8

65.8

0.73

31,006

0.77

54.9

58.1

0.59

Droplet

20

0.64

93.3

100.0

0.96

20

0.23

86.7

60.0

0.67

Back to article page