Augmenting the National Institutes of Health Chest Radiograph Dataset with Expert Annotations of Possible Pneumonia | JoVE Visualize

Augmenting the National Institutes of Health Chest Radiograph Dataset with Expert Annotations of Possible Pneumonia

George Shih¹, Carol C Wu¹, Safwan S Halabi¹

¹Department of Radiology, Weill Cornell Medical College, 525 E 68th St, Box 141, New York, NY 10065 (G.S.); Department of Diagnostic Radiology, University of Texas MD Anderson Cancer Center, Houston, Tex (C.C.W., M.C.B.G.); Department of Radiology, Stanford University School of Medicine, Stanford, Calif (S.S.H.); Department of Radiology and Biomedical Imaging, University of California-San Francisco, San Francisco, Calif (M.D.K.); Department of Radiology, The Ohio State University Wexner Medical Center, Columbus, Ohio (L.M.P.); Department of Radiology, Hospital of the University of Pennsylvania, Philadelphia, Pa (T.S.C., M.G.); Department of Radiology, Amita Health, Chicago, Ill (A. Sharma); Department of Radiology, Rutgers Robert Wood Johnson Medical School, New Brunswick, NJ (J.K.A.); Department of Radiology, University of Arizona College of Medicine, Tucson, Ariz (V.A.); Department of Radiology, Beth Israel Deaconess Medical Center, Boston, Mass (R.R.G.); Department of Radiology, University of Kentucky College of Medicine, Lexington, Ky (S.H.); Department of Diagnostic Radiology and Nuclear Medicine, University of Maryland School of Medicine, Baltimore, Md (J.J.); Department of Radiology, University of Iowa Carver College of Medicine, Iowa City, Iowa (A.L.); Department of Diagnostic Radiology, Rush University Medical Center, Chicago, Ill (P.N.S.); Department of Radiology, University of Michigan Health System, Ann Arbor, Mich (D.V.); Department of Radiology, Stony Brook School of Medicine, Stony Brook, NY (K.Y.); and MD.ai, New York, NY (A. Stein).

Radiology. Artificial Intelligence|May 3, 2021

View abstract on PubMed

Summary

This dataset provides bounding box annotations for pulmonary opacity on chest radiographs. These annotations aid machine learning models in identifying potential pneumonia in clinical settings.

Area of Science:

Medical Imaging
Machine Learning
Radiology

Background:

Pulmonary opacity on chest radiographs can indicate pneumonia.
Accurate identification of pneumonia is crucial for timely treatment.
Machine learning requires well-annotated datasets for training.

Purpose of the Study:

To create a dataset for machine learning model development.
To provide bounding box annotations for pulmonary opacities.
To facilitate the detection of pneumonia using artificial intelligence.

Main Methods:

Chest radiographs were collected.
Pulmonary opacities were identified.
Bounding box annotations were generated for machine learning.

Main Results:

A dataset with bounding box annotations for pulmonary opacity was created.
The dataset is suitable for training machine learning algorithms.
Annotations highlight regions potentially indicative of pneumonia.

Conclusions:

This annotated dataset supports the advancement of AI-driven pneumonia detection.
The dataset enables the development of more accurate machine learning models for chest radiograph analysis.
It serves as a valuable resource for research in medical imaging and artificial intelligence.