Skip to main content

Table 3 Test performance from the proposed scheme on TCGA and MICCAI datasets with and without adding GAN-augmented images in the training

From: Deep semi-supervised learning for brain tumor classification

Dataset

GAN

Accuracy(|σ|)

Sensitivity(|σ|)

Specificity(|σ|)

TCGA

Without

85.30(4.43)

73.75(9.27)

90.91(5.25)

 

With

86.53(4.24)

73.75(8.15)

92.73(3.45)

MICCAI

Without

89.53(1.42)

78.26(4.35)

93.65(2.75)

 

With

90.70(1.42)

84.35(6.59)

93.01(1.42)

  1. Results are shown in mean value (standard deviations |σ|) over 5 runs, all values are in percentage %. The best results in accuracy/sensitivity/specificity are marked in boldface
Back to article page

BMC Medical Imaging

ISSN: 1471-2342

Contact us