Table 2 MEASURES OF PERFORMANCE FOR THE SIX MODELS STUDIED IN THE RESEARCH.

From: Vision transformer and explainable transfer learning models for auto detection of kidney cyst, stone and tumor from CT-radiography

Models

Accuracy

Class

Precision (PPV)

Recall (Sensitivity)

F1 Score

AUC

EANet

77.02%

Cyst

0.593

1

0.745

0.98

Normal

0.896

0.848

0.871

0.98

Stone

0.845

0.495

0.624

0.91

Tumor

0.93

0.777

0.847

0.97

Swin Transformers

99.30%

Cyst

0.996

0.996

0.996

0.99993

Normal

0.996

0.981

0.988

0.9998

Stone

0.981

0.989

0.985

0.99975

Tumor

0.993

1

0.996

1

CCT

96.54%

Cyst

0.968

0.923

0.945

0.99605

Normal

0.989

0.975

0.982

0.99841

Stone

0.94

1

0.969

0.99924

Tumor

0.964

0.964

0.964

0.99723

VGG16

98.20%

Cyst

0.996

0.968

0.982

0.99856

Normal

0.985

0.973

0.979

0.99844

Stone

0.966

0.988

0.977

0.99908

Tumor

0.982

0.996

0.989

0.99902

Inception v3

61.60%

Cyst

0.645

0.826

0.724

0.92689

Normal

0.584

0.898

0.708

0.90642

Stone

0.568

0.462

0.509

0.78185

Tumor

0.76

0.295

0.425

0.8029

Resnet50

73.80%

Cyst

0.735

0.641

0.685

0.90721

Normal

0.77

0.79

0.78

0.95069

Stone

0.745

0.692

0.717

0.9314

Tumor

0.706

0.827

0.762

0.94447

Back to article page