A single-breath–hold magnetic resonance imaging protocol accelerated with artificial intelligence–assisted compressed sensing may detect clinically relevant pulmonary nodules with high accuracy, according to a prospective study.
In a single-center prospective study, researchers recruited 97 asymptomatic adult patients who underwent whole-body magnetic resonance imaging (MRI) screening between June 2024 and June 2025 and also underwent same-day noncontrast photon-counting detector computed tomography (CT) within 24 hours, which served as the reference standard. They evaluated the diagnostic performance of a single-breath–hold, artificial intelligence (AI)-assisted compressed-sensing three-dimensional T1-weighted fast field echo sequence for pulmonary nodule detection compared with CT, with secondary analyses of Lung-RADS classification, nodule size agreement, and detection according to nodule size and composition.
CT identified 118 pulmonary nodules, including 73 solid nodules. Overall, MRI detected 78 nodules, with just one false-positive finding. For solid and part-solid nodules, MRI achieved 83% sensitivity and 87% accuracy. Performance improved for clinically relevant solid nodules measuring at least 4 mm, reaching 98% sensitivity and 98% accuracy. Diagnostic performance increased further for nodules measuring at least 6 mm and 8 mm.
The imaging modalities demonstrated complete agreement for Lung-RADS categories 4A and 4B, with all suspicious lesions identified on MRI confirmed by CT. Agreement for nodule size was also excellent, although MRI had an average nodule diameter underestimation of about 1 mm. No clinically relevant suspicious nodules were missed in this cohort.
Diagnostic performance varied according to nodule size and composition. MRI had reduced sensitivity for nodules smaller than 4 mm and calcified nodules. More than 50% of the missed lesions were small calcified nodules, while most of the remaining missed lesions were very small solid nodules. The researchers noted that the findings reflected the technical limitations of MRI for very small lesions rather than clinically important screening abnormalities.
The study had several limitations. It was conducted at a single center and included a relatively small cohort of asymptomatic patients, limiting generalizability. The protocol also showed reduced sensitivity for nodules smaller than 4 mm, particularly calcified nodules, and requires validation in larger screening cohorts and across multiple imaging platforms prior to broader clinical adoption.
The findings suggested that a rapid, single-breath–hold MRI examination may offer a practical radiation-free approach in detecting clinically relevant pulmonary nodules while maintaining close agreement with CT for screening classification. Additional studies in larger and higher-risk populations are needed to confirm its role in lung cancer screening.
“[L]ung MRI represents a promising alternative to CT for lung nodule screening, with potential applicability both for population-wide lung cancer screening and selected subgroups such as younger individuals, patients requiring repeated follow-up examinations, or those who decline CT,” wrote lead study author Anna Palmisano, of the Advanced Imaging for Personalized Medicine Unit at the Experimental Imaging Center at the IRCCS San Raffaele Hospital in Italy, and colleagues.
The study was partially supported by the Piano Nazionale di Ripresa e Resilienza. Co–study author Giulio Ferrazzi reported being a researcher at Philips. The study authors reported no other conflicts of interest.
Source: European Radiology
