Prognosis Assessment in Spontaneous (non-traumatic) Intracerebral Hemorrhage with Artificial Intelligence-Assisted Hemorrhage Volume Analysis

Objective: Spontaneous (non-traumatic) intracerebral hemorrhage (sICH) is associated with high mortality and morbidity rates. With the increasing incidence of sICH, hemorrhage volume and hemorrhage location on brain computed tomography (CT) are important in determining prognosis. CT scans obtained from patients with sICH are reported using artificial intelligence (AI)-assisted programs. These programs provide data on the type, volume, and location of bleeding. In this study, we aimed to investigate the reliability of AI-assisted hemorrhage volume measurement, the effect of measured volume on QTc, and the contribution of these parameters to pre-dicting mortality in patients with sICH.
Material and Methods: The study was designed as a retrospective, single-center cohort study. Hemorrhage volumes on CT images were calculated using AI al-gorithms from Hevi AI. QTc values were calculated using the Bazett formula, and statistical analyses were conducted by grouping patients according to 1-week and 1-month mortality.
Results: Eighty-five patients diagnosed with sICH were included in the study. The mean age of the patients was 62.9±14.6 years. No significant association was ob-served between age and 1-month mortality (p=0.890). Large hemorrhage volume, low Glasgow Coma Scale (GCS) score, and prolonged QTc duration were significantly associated with 1-week mortality (p<0.001). Hemorrhage volume showed a moderate-to-high significant negative correlation with GCS (r=-0.755, p<0.001) and a moderately significant positive correlation with QTc (r=0.477, p<0.001). In the Cox regression analysis performed to determine the effect of risk factors on mortality, large hemorrhage volume and low GCS level increased the probability of 1-week mortality (p=0.001, hazard ratio=1.018, confidence interval [CI]=1.008–1.029; and p=0.020, HR=0.852, CI=0.745–0.975, respectively).
Conclusion: AI-assisted measurement of large hemorrhage volume and low GCS appear to be important prognostic indicators, particularly regarding 1-week mortality.

Keywords: Artificial intelligence; hemorrhage volume; mortality; QTc interval; spontaneous (non-traumatic) intracerebral hemorrhage