MRI based volumetric lung nodule assessment - a comparison to computed tomography

Previous studies have demonstrated that nodule volumetry allows for the deduction of imaging-based biomarkers such as volume doubling time, enabling superior discrimination between benign and malignant lesions compared to 2D-based morphological characteristics. The study aimed to assess the feasibility and accuracy of in-vivo magnetic resonance imaging (MRI)-based volumetric assessment of lung nodules larger than 6 mm, in comparison to the current gold standard, CT.

This study involved a subgroup analysis of 233 participants from a prospective, single-center lung cancer screening program using CT and MRI. Patients were included if foci ≥6 mm were detected in CT during the initial screening round, resulting in 23 participants with 47 pulmonary nodules. MRI was performed using a 1.5 Tesla unit with a transverse T2-weighted MultiVane XD imaging technique, while low-dose CT (LDCT) was performed on a 128-slice spiral CT scanner. Volumetric nodule assessment was conducted using a computer-aided diagnosis system, with images reviewed by two experienced radiologists. Statistical analysis included regression analysis, Bland-Altman analysis, and calculation of the interclass correlation coefficient (ICC) to assess correlation and reproducibility.

Comparison of MRI-based volumetric assessment with LDCT as the reference standard revealed a mean nodule volume of 1.1343 ± 3.1204 cm³ for MRI versus 1.2197 ± 3.496 cm³ for LDCT (p = 0.203). Regression analysis demonstrated a strong linear relationship between the modalities (r ² = 0.981, p < 0.001), consistently observed even for nodules <5 cm³ (r ² = 0.755, p < 0.001). Bland-Altman analysis indicated no significant systematic bias in nodule volume measurements between MRI and CT, with a mean difference of 0.12 cm³ and narrow 95% confidence intervals (-6.852 to 6.854 cm³). Intra-reader reproducibility for CT-based volumetry was excellent (ICC = 0.9984), while MRI-based measurements showed good reproducibility (ICC = 0.7737). Inter-reader reproducibility was high for CT (ICC = 0.995) and moderate for MRI (ICC = 0.7135).

This study demonstrates that MRI-based volumetry of lung nodules ≥6 mm is feasible and accurate, showing comparable precision to CT with minimal bias in volume measurements, and highlights the potential of MRI as a radiation-free alternative for lung nodule follow-up and screening.