Automated CT segmentation of kidneys successfully quantifies changes in total kidney volume in ICU patients, a retrospective cohort study

Abstract Background Kidney function is associated with kidney volume. This study aims to explore automated segmentation for measuring total kidney volume (TKV) and to analyse the association between (changes in) TKV and acute kidney injury (AKI) incidence and/or severity in Intensive Care Unit (ICU) patients. Methods Patients were included in this retrospective pilot cohort study when at least two abdominal Computed Tomography (CT) scans were performed during ICU admission. If available, CT scans made before the ICU admission were included as a baseline scan. TKV was measured by automated segmentation of both kidneys using Data Analysis Facilitation Suite (DAFS, Voronoi Analytics Incorporated). All segmentations were visually checked and manually adjusted when necessary. ΔTKV was calculated between baseline CT and CT1 (ΔTKVCT1–baseline) and CT1 and CT2 (ΔTKVCT2–CT1). Primary outcomes were differences in kidney volume before and after manual correction and AKI incidence and severity, per the Kidney Disease: Improving Global Outcomes (KDIGO) guidelines, on the day of scanning. Results Twenty-six patients were included, of whom eighteen developed AKI during ICU admission. Analysis showed no significant differences in volumes before and after manual correction of the automated segmentations. TKV was not associated with AKI incidence or severity. Longitudinal intraindividual changes in TKV were observed. Median ΔTKVCT1–baseline was statistically significantly different for AKI versus non-AKI patients (−22 cm3 (−49–9) versus 42 cm3 (23–43), p = 0.03) and for different KDIGO stages. Conclusion This study demonstrates the possibility of measuring TKV on CT in ICU patients using automated segmentation. Longitudinal intraindividual changes in TKV were observed, however, no clear association between TKV and AKI was found. Clinical trial number Not applicable.