Performance of multi-planar volume rendering versus maximum intensity projection and minimum intensity projection on pulmonary nodule detection.

Background: Pulmonary nodule detection is critical for the early diagnosis of lung cancer. However, the increasing volume of thin-section computed tomography (CT) data challenges radiologists' accuracy and efficiency. While post-processing techniques like maximum intensity projection (MIP) and minimum intensity projection (MinIP) are used to improve detection, they have limitations in visualizing nodules of varying densities and sizes. Multiplanar volume rendering (MPVR), an advanced three-dimensional (3D) reconstruction technique, may overcome these drawbacks. This study aimed to systematically compare the diagnostic efficacy of MPVR with conventional MIP and MinIP for detecting pulmonary nodules across different densities and sizes. Methods: This retrospective study analyzed thin-section CT images from 183 patients diagnosed with pulmonary nodules between January 2023 and October 2023. MPVR, MIP, and MinIP images were independently reviewed by two radiologists, with nodule detection rates compared across density [solid nodules (SNs), part-SNs (PSNs), ground-glass nodules (GGNs)] and size categories (<5 mm, 5-10 mm, >10 mm). Statistical analysis including Wilcoxon and Bayesian Wilcoxon signed-rank tests, and interobserver agreement was assessed using kappa statistics. Results: A total of 1,018 SNs, 146 PSNs, and 583 GGNs were detected across all imaging methods. MPVR demonstrated superior detection rates, identifying 91% of SNs, 85% of PSNs, and 71% of GGNs, compared to 74%, 47%, and 58% with MIP and 51%, 48%, and 86% with MinIP, respectively. For nodules smaller than 5 mm, MPVR detected 87%, significantly higher than 69% with MIP and 60% with MinIP (P<0.05 and Bayes factor >3). MPVR showed excellent interobserver agreement (κ=0.913), which was comparable to MIP (κ=0.949) and MinIP (κ=0.972). The use of optimized threshold settings and pseudo-color visualization in MPVR facilitated improved differentiation between nodules and surrounding tissues. Conclusions: MPVR demonstrated superior performance over MIP and MinIP in detecting most types and sizes of pulmonary nodules; however, MinIP was found to be more effective for GGNs. These findings highlight MPVR's potential to improve early lung cancer diagnosis, streamline radiological workflows, and reduce diagnostic variability. Future studies integrating MPVR with artificial intelligence-based detection tools are warranted to further validate its clinical utility.