Atlas
Sign in
Models / medical-imaging / chest-x-ray-basic / benchmarks

medical-imaging/chest-x-ray-basic

Visibility: PUBLIC
Model cardFilesBenchmarks
Group by

Chest X-ray findings (NIH ChestX-ray14) medical-imaging/chest-xray-findings-classification

Drawn from: NIH-Chest-Xray-14

atlas-full-v2

Metric Value 95% CI n Rank Best in catalogue Source Overlap Board
Mean ROC-AUC, 5 findings (logistic probe) ↑ higher is better · probe
By finding: 5 levels
findingValue [95% CI]n
Atelectasis0.638 [0.585, 0.688]1013
Cardiomegaly0.552 [0.459, 0.653]1013
Consolidation0.591 [0.522, 0.657]1013
Edema0.767 [0.666, 0.851]1013
Pleural Effusion0.694 [0.655, 0.734]1013

levels share items; n is per level

 0.649 [0.614, 0.684] 1013 2 / 5 0.689 reproduced
overlap

Train/test overlap declared by the evaluator: medical-imaging/nih-chest-xray-14

 Board
Mean average precision, 5 findings (logistic probe) ↑ higher is better · probe · secondary
By finding: 5 levels
findingValue [95% CI]n
Atelectasis0.221 [0.178, 0.289]1013
Cardiomegaly0.042 [0.028, 0.070]1013
Consolidation0.108 [0.080, 0.153]1013
Edema0.092 [0.057, 0.168]1013
Pleural Effusion0.303 [0.253, 0.370]1013

levels share items; n is per level

 0.153 [0.139, 0.182] 1013 — 0.203 reproduced
overlap

Train/test overlap declared by the evaluator: medical-imaging/nih-chest-xray-14

 Board

Pediatric pneumonia detection (chest X-ray) medical-imaging/pediatric-pneumonia-detection

Drawn from: chest-xray-pneumonia

atlas-full-v1

Metric Value 95% CI n Rank Best in catalogue Source Overlap Board
ROC-AUC (logistic probe) ↑ higher is better · probe 0.958 [0.943, 0.972] 624 4 / 5 0.992 reproduced none declared Board
Accuracy (logistic probe) ↑ higher is better · probe · secondary 0.851 [0.825, 0.875] 624 — 0.962 reproduced none declared Board
Sensitivity, pneumonia (logistic probe) ↑ higher is better · probe · secondary 0.992 [0.982, 1.000] 624 — 0.997 reproduced none declared Board
Specificity, pneumonia (logistic probe) ↑ higher is better · probe · secondary 0.615 [0.547, 0.675] 624 — 0.902 reproduced none declared Board
Average precision (logistic probe) ↑ higher is better · probe · secondary 0.971 [0.957, 0.982] 624 — 0.994 reproduced none declared Board