Work

3.5/6SH: Multicenter Real-World Derivation and External Validation of a CT-Based Formula for Chronic Subdural Hematoma Volume Estimation

Year: 2026

Type: article

Abstract: analysis of a multicenter, real-world study compared seven simplified computed tomography-based formulas with volumetric measurements obtained using three-dimensional (3D) Slicer in a derivation cohort of 129 patients with cSDH. We further evaluated the newly proposed 3.5/6SH formula in an independent validation cohort of 60 additional patients. Accuracy, agreement, and efficiency were evaluated using estimation error, percentage deviation, intraclass correlation coefficients (ICCs), and Bland-Altman (BA) analysis. In the derivation cohort, the 3.5/6SH formula yielded the lowest median volume deviation (-0.82 mL, interquartile range [IQR] 13.12 mL), the smallest percentage deviation (-1.00%, IQR 17.08%), and the highest ICC (0.973), indicating high accuracy and close agreement with the 3D Slicer reference standard. In contrast, the widely used 1/2ABC formula exhibited a larger negative percentage deviation (-4.64%) and a lower ICC (0.910). BA analysis further confirmed that 3.5/6SH had the narrowest limits of agreement and a clinically acceptable maximum absolute percentage deviation (29.33%), whereas the maximum deviation for 1/2ABC (65.04%) exceeded the clinical acceptability threshold (33%). In an independent validation cohort of 60 patients, 3.5/6SH maintained a small bias (median deviation -1.29 mL; percentage deviation -1.27%) and excellent agreement with 3D Slicer (ICC = 0.979). Inter- and intrarater ICCs for 3.5/6SH measurements in the validation cohort were 0.994 and 0.998, respectively, supporting excellent reproducibility. The average calculation time was 77.95 sec for 3.5/6SH-longer than for 1/2ABC (35.95 sec), but substantially shorter than for 3D Slicer (19.04 min). The 3.5/6SH formula showed high accuracy and agreement with 3D Slicer across all metrics and offers a reliable, simplified method for bedside cSDH volume estimation that may facilitate more consistent volume assessment in multicenter research and clinical practice, particularly for more sensitive monitoring of hematoma changes in resource-limited environments.

Source: Journal of Neurotrauma

DOI: 10.1177/08977151251414140