Deep learning SR-UTE MRI significantly improves detection of small pulmonary nodules in emphysema

In a prospective study of 127 patients with emphysema, deep learning-based SR-UTE MRI detected 81.8% of nodules <6 mm vs 54.5% with conventional UTE (p<0.05), with superior size agreement (ICC 0.89-0.93) and longitudinal reproducibility (ICC 0.90 vs 0.82) versus LDCT reference.

• Prospective study of 127 patients with emphysema and pulmonary nodules; LDCT served as reference standard.
• For ground-glass nodules in severe emphysema, SR-UTE detection rate was 66.7% vs 33.3% for UTE (improvement of 33.4 percentage points).
• Both methods achieved 100% sensitivity for nodules >8 mm; SR-UTE also provided higher SNR and CNR in severe emphysema (p<0.05).

Abstract does not report whether the study was single-center or multi-center, nor if the deep learning model underwent external validation.

Automated summary

RadPigeon summaries are original and for information only. They are not clinical advice.