[1]	[1]	[2]	[3]	[4]	[7]	[8]	[9]	[10]
SPLIT BY	patient	image	image	patient	image	image	image	patient	patient
OFFICIAL SPLIT	Yes	No	No	No	No	No	No	Yes	Yes
Atelectasis	0.7003	0.72	0.81	0.772	0.80	0.76	0.853	0.767	0.733
Cardiomegaly	0.8100	0.81	0.904	0.9248	0.81	0.91	0.939	0.883	0.858
Effusion	0.7585	0.78	0.859	0.8638	0.87	0.86	0.903	0.828	0.806
Infiltration	0.6614	0.61	0.695	0.7345	0.70	0.69	0.754	0.709	0.675
Mass	0.6933	0.71	0.792	0.8676	0.83	0.78	0.902	0.821	0.727
Nodule	0.6687	0.67	0.717	0.7802	0.75	0.70	0.828	0.758	0.778
Pneumonia	0.6580	0.63	0.713	0.7680	0.67	0.71	0.774	0.731	0.690
Pneumothorax	0.7993	0.81	0.841	0.8887	0.87	0.86	0.921	0.846	0.805
Consolidation	0.7032	0.71	0.788	0.7901	0.80	0.78	0.842	0.745	0.717
Edema	0.8052	0.83	0.882	0.8878	0.88	0.89	0.924	0.835	0.806
Emphysema	0.8330	0.81	0.829	0.9371	0.91	0.90	0.932	0.895	0.842
Fibrosis	0.7859	0.77	0.767	0.8047	0.78	0.76	0.864	0.818	0.757
Pleural Thickening	0.6835	0.71	0.765	0.8062	0.79	0.77	0.837	0.761	0.724
Hernia	0.8717	0.77	0.914	0.9164	0.77	0.90	0.921	0.896	0.824

Split by image: This repo contains the splits of train, valid and test.

Split by patient: Please be aware that the official splits by patient are only recently available here

Please contribute to the following list:

[1] ChestX-ray8: Hospital-scale Chest X-ray Database and Benchmarks on Weakly-Supervised Classification and Localization of Common Thorax Diseases

[2] Learning to diagnose from scratch by exploiting dependencies among labels

[3] CheXNet: Radiologist-Level Pneumonia Detection on Chest X-Rays with Deep Learning

[4] Thoracic Disease Identification and Localization with Limited Supervision

[5] Learning to detect chest radiographs containing lung nodules using visual attention networks(Private dataset)

[6] TieNet: Text-Image Embedding Network for Common Thorax Disease Classification and Reporting in Chest X-rays (different tasks, no improvement on using only images)

[7] Boosted Cascaded Convnets for Multilabel Classification of Thoracic Diseases in Chest Radiographs

[8] Diagnose like a Radiologist: Attention Guided Convolutional Neural Network for Thorax Disease Classification

[9] Learning to recognize Abnormalities in Chest X-Rays with Location-Aware Dense Networks (SOTA achived with extra PLCO dataset)

[10] Weakly Supervised Medical Diagnosis and Localization from Multiple Resolutions (SOTA without using extra data)