Table 7 Effect of each module on the model performance.

From: Deep multi-task learning framework for gastrointestinal lesion-aided diagnosis and severity estimation

Model variant module

Accuracy

Precision

Recall

F1 score

Without EMA

0.939

0.933

0.926

0.930

Without CViT

0.931

0.918

0.911

0.914

GTCAD (Full)

0.965

0.960

0.952

0.958

Back to article page