Sex-, Age-, and Projection-Related Variability in Pneumothorax and Pulmonary Mass Detection Frequencies on Chest...

Area of Science:

• Radiology and Medical Imaging
• Epidemiological Research
• Data Science in Healthcare

Background:

• Chest radiography is a primary tool for detecting thoracic abnormalities like pneumothorax and pulmonary masses.
• Detection rates of these findings can differ based on patient demographics and imaging techniques.
• Understanding variability is key for accurate interpretation of large imaging datasets and epidemiological studies.

Purpose of the Study:

• To investigate sex-, age-, and projection-related variations in pneumothorax and pulmonary mass detection frequencies.
• To analyze these differences within a large chest radiograph dataset (NIH ChestX-ray14).

Main Methods:

• Retrospective analysis of 112,120 frontal chest radiographs from 30,805 adult patients.
• Inclusion of anteroposterior (AP) and posteroanterior (PA) projections; exclusion of lateral projections and patients under 18.
• Utilized natural language processing (NLP) for annotation extraction and statistical analyses including chi-square tests and logistic regression (GEE).

Main Results:

• Pneumothorax detection was higher in females aged 41-80; pulmonary masses were more frequent in males aged 21-40.
• Females were older and more likely to have PA imaging; AP projection was linked to lower pneumothorax detection.
• Patient-level analyses confirmed image-level findings, highlighting demographic and projection influences on detection frequencies.

Conclusions:

• Significant demographic (sex, age) and technical (projection) variability exists in chest radiograph annotations.
• These variations necessitate consideration for accurate interpretation of large datasets and reliable imaging-based research.
• Findings represent dataset associations, not disease prevalence, emphasizing the need for context in interpretation.