Table 2 Evaluation results on TMAs dataset as Mean ± std and the bold font is the best scores.

From: Enhanced hierarchical attention mechanism for mixed MIL in automatic Gleason grading and scoring

Method

Per-class k-score

K-scores

AUC

Benign

Grade3

Grsde4

Grade5

ABMIL

–

–

–

–

–

0.6672 ± 0.0237

CLAM

0.6651 ± 0.0034

0.3459 ± 0.0024

0.2835 ± 0.0007

0.3905 ± 0.0017

0.2720 ± 0.0023

0.6563 ± 0.0548

DSMIL

0.8395 ± 0.0027

0.1756 ± 0.0056

0.0416 ± 0.0015

0.3475 ± 0.0025

0.1317 ± 0.0052

0.6137 ± 0.0452

ATMIL

–

–

–

–

–

0.7564 ± 0.0142

TransMIL

0.7107 ± 0.0076

0.7921 ± 0.0017

0.4853 ± 0.0024

0.0433 ± 0.0121

0.5153 ± 0.0027

0.7198 ± 0.0852

Mixed MIL

0.6793 ± 0.0065

0.5869 ± 0.0037

0.5497 ± 0.0018

0.2171 ± 0.0012

0.4605 ± 0.0014

0.8109 ± 0.0969

Ours

0.7786 ± 0.0162

0.7924 ± 0.0024

0.6206 ± 0.0003

0.7825 ± 0.0035

0.7779 ± 0.0045

0.8889 ± 0.0257

Back to article page