Automated quantitative chest CT for mortality prediction in COVID-19 patients in a resource-limited emergency setting...

To assess the prognostic value of automated quantitative chest CT metrics in predicting in-hospital mortality among patients with COVID-19 pneumonia admitted through the emergency department in a resource-limited setting in Gaza.

This retrospective study included 300 adult patients with RT-PCR-confirmed COVID-19 pneumonia who underwent non-contrast chest CT upon hospital admission. Automated quantitative lung metrics were derived using LungCTAnalyzer, an open-source 3D Slicer extension. Metrics included functional lung volume, affected lung volume, and the COVID-Q index (affected-to-functional lung ratio). Patients were stratified by survival status, and outcomes were analyzed using ROC curves, Kaplan-Meier survival analysis, and log-rank testing.

Among the cohort, 112 patients (37.3%) died during hospitalization. Non-survivors were older and more likely to require advanced respiratory support (p < 0.001). Quantitative CT analysis revealed significantly reduced functional lung volume (47.2% vs. 73.9%) and increased affected lung volume (52.8% vs. 26.1%) in non-survivors (p < 0.001). The COVID-Q index was markedly higher in the deceased group. ROC analysis showed good predictive performance for total affected lung volume (AUC = 0.756; 95% CI: 0.696-0.815), with an optimal threshold of approximately 42%. Right lung involvement was associated with the poorest survival outcomes (log-rank = 67.6, p < 0.001).

Automated quantitative chest CT provides objective, reproducible metrics for early mortality risk stratification in COVID-19 pneumonia. The use of open-source tools like LungCTAnalyzer may assist emergency radiologists in prioritizing care in resource-constrained and conflict-affected healthcare systems.