Variability Among Radiologists in Interpreting Low-Dose CT Lung Cancer Screens

Area of Science:

• Radiology
• Oncology
• Medical Imaging Analysis

Background:

• Low-dose CT (LDCT) screening for lung cancer reduces mortality but faces challenges with false positive results.
• Radiologist interpretation variability may contribute to elevated false positive rates (FPRs) in LDCT scans.

Purpose of the Study:

• To evaluate the variability in radiologists' false positive rates (FPRs) during lung cancer screening (LCS).
• To assess the relationship between radiologist FPR and sensitivity in LDCT screening.

Main Methods:

• Utilized a merged database of the Lung Cancer Screening Registry (LCSR) and Medicare files.
• Included radiologists with at least 50 baseline or post-baseline LDCT screens.
• Defined sensitivity and FPR based on cancer diagnosis and calculated distributions using observed data and mixed-effects models.

Main Results:

• High variability in median FPRs was observed for both baseline (15.9%-16.3%) and post-baseline (8.5%-8.8%) LDCT screens.
• Radiologists with higher FPR quartiles demonstrated significantly increased sensitivity for both baseline and post-baseline screens.
• For baseline screens, the upper two FPR quartiles showed higher sensitivity (94.4%) compared to the lower two (89.6%).

Conclusions:

• Significant variability exists in radiologists' false positive rates in LDCT lung cancer screening.
• Increased false positive rates are associated with higher sensitivity in detecting lung cancer.