Skip to main content

Table 3 Results of evaluation using independent test data with a size of 21 tissue samples patches equally distributed at the checkpoint after 254 epochs with lowest loss of 0.1321 as shown in Fig. 6. Values are reported as class-dependent classification scores (Acc = accuracy, Spec = specificity, Sens = sensitivity)

From: Hyperspectral imaging and artificial intelligence to detect oral malignancy – part 1 - automated tissue classification of oral muscle, fat and mucosa using a light-weight 6-layer deep neural network

Class

Samples

Acc

Spec

Sens

Fat

7

0.95

1.00

0.86

Muscle

7

0.86

0.79

1.00

Mucosa

7

0.81

0.93

0.57