Objective:

To evaluate the diagnostic performance of an AI-assisted MRI protocol for detecting pulmonary nodules compared to CT.

Approach:

• Study Design: A single-center prospective study involving 97 asymptomatic adult patients who underwent whole-body MRI and same-day CT.
• Imaging Protocol: Utilized a single-breath-hold, AI-assisted compressed-sensing three-dimensional T1-weighted fast field echo sequence.

Key Findings:

• CT identified 118 pulmonary nodules, while MRI detected 78 with one false-positive.
• MRI achieved 83% sensitivity and 87% accuracy for solid and part-solid nodules.
• For solid nodules ≥4 mm, MRI sensitivity and accuracy reached 98%.
• No clinically relevant suspicious nodules were missed in the cohort.
• MRI showed reduced sensitivity for nodules <4 mm and calcified nodules.

Interpretation:

Limitations:

• Conducted at a single center with a small cohort, limiting generalizability.
• Reduced sensitivity for nodules <4 mm, particularly calcified nodules.
• Requires validation in larger screening cohorts and across multiple imaging platforms.

Conclusion:

Sources:

• European Radiology