Table 3 Performance comparison between LiAIDS and radiologists on detection and diagnosis of FLLs

From: A multicenter clinical AI system study for detection and diagnosis of focal liver lesions

Approach

Recall of detection

Accuracy

Benign lesions

Malignant lesions

Sen

Spe

F1-score

Sen

Spe

F1-score

Junior1

0.766

0.727

0.767

0.361

0.830

0.361

0.767

0.230

Junior2

0.883

0.810

0.857

0.452

0.890

0.452

0.857

0.360

Junior3

0.796

0.767

0.804

0.494

0.860

0.494

0.804

0.330

Senior1

0.911

0.905

0.937

0.663

0.950

0.663

0.937

0.620

Senior2

0.911

0.870

0.894

0.688

0.920

0.688

0.894

0.550

Senior3

0.962

0.907

0.931

0.729

0.950

0.729

0.931

0.650

Junior1+LiAIDS

0.97

0.893

0.899

0.854

0.936

0.854

0.899

0.680

Junior2+LiAIDS

0.954

0.907

0.933

0.734

0.946

0.734

0.933

0.671

Junior3+LiAIDS

0.965

0.904

0.931

0.728

0.944

0.728

0.931

0.667

Senior1+LiAIDS

0.975

0.932

0.957

0.771

0.961

0.771

0.957

0.753

Senior2+LiAIDS

0.969

0.913

0.946

0.695

0.950

0.695

0.946

0.679

Senior3+LiAIDS

0.972

0.927

0.953

0.750

0.958

0.750

0.953

0.727

LiAIDS

0.965

0.900

0.900

0.900

0.940

0.900

0.900

0.692

  1. LiAIDS Liver Artificial Intelligence Diagnosis System, Spe Specificity, Sen Sensitivity.