A newly developed chest CT scoring system, the RACOON Viral Pneumonia Score, demonstrated excellent inter-reader reliability and greater diagnostic detail than the Pan Score when evaluating pneumonia-related lung findings, including those caused by COVID-19. The multicenter proof-of-concept study was based on 8,525 observations and aimed to address the limitations of one-dimensional CT scoring tools.
Eight experienced radiologists performed 7,800 RVPS readings across 624 scorings of 39 CT scans. Inter-reader agreement was high for acute infection findings: Kendall’s W was 0.81 for ground-glass opacities, 0.86 for interstitial thickening, and 0.87 for consolidations. Agreement for postinfectious findings was moderate: W 0.55 for linear opacities and W 0.69 for fibrotic-like changes; and summed categories showed strong reliability: W 0.95 for infectious findings and W 0.79 for postinfectious findings.
In COVID-19 cases, median RACOON Viral Pneumonia Score (RVPS) scores aligned with clinical disease stage: 13 in stage 1, 18 in stage 2, 24 in stage 3, 18 in stage 4, and 10 in stage 5. This progression illustrated the shift from ground-glass patterns in early stages to fibrotic changes in later phases that could not be captured by Pan Score values alone. For example, an identical Pan Score could describe either acute infection or irreversible fibrosis, whereas the RVPS differentiated between them through its subscore structure.
The RVPS also identified pneumonia-specific imaging signatures: Pneumocystis jirovecii pneumonia was marked by ground-glass opacities and interstitial thickening, while bacterial superinfection of respiratory syncytial virus was noted as ground-glass opacities with dense consolidation. These distinctions were not visible using Pan Score data.
Pan Score is widely used as a one-dimensional visual CT scoring system that was developed during the early stages of the COVID-19 pandemic. Unlike the Pan Score, which estimates total lung involvement using a single number per lobe, the RVPS evaluates five defined CT patterns per lung region: ground-glass opacities, ground-glass opacities with interstitial thickening, consolidations, linear opacities and reticulations, and fibrotic-like changes. Each finding is scored from 0 to 5 depending on the percentage of lobe affected, with a maximum possible total of 52.
The Pan Score, while simple and fast to apply (70 ± 33 seconds per scan), does not differentiate among types of pneumonia-related lung abnormalities. In contrast, the RVPS requires more time per scan (129 ± 40 seconds) but provides nuanced, pattern-specific data that allows radiologists to characterize pneumonia stage and progression more clearly.
“With the RVPS, we offer a comprehensive scoring method for standardized, long-term assessment of infectious and post-infectious lung diseases,” wrote Thorsten Persigehl, MD, of the Institute for Diagnostic and Interventional Radiology, University Hospital Cologne, University of Cologne in Germany, with colleagues.
The study included 21 adult patients (mean age = 57.7 years; 16 males), 12 of whom had COVID-19 pneumonia. The others were diagnosed with influenza, Pneumocystis jirovecii, Streptococcus pneumoniae, or respiratory syncytial virus. Each underwent a minimum of three CT scans, which were scored using both the RVPS and Pan Score.
The authors noted that RVPS is machine-readable and compatible with artificial intelligence integration, and therefore may be able to support longitudinal monitoring and clinical trials. Its modular design may allow future adaptation to new or emerging pulmonary pathogens. Although the study was limited by a small, nonrandom patient population, it represents an initial validation of RVPS performance. A multicenter study across the RACOON network is currently underway to expand evaluation.
The authors reported no conflicts of interest.
Source: frontiers