Accuracy of automated intracerebral hemorrhage volume measurement on non-contrast computed tomography: a Swedish Stroke Register cohort study

Abstract

Purpose Hematoma volume is the strongest predictor of patient outcome after intracerebral hemorrhage (ICH). The aim of this study was to validate novel fully automated software for quantification of ICH volume on non-contrast computed tomography (CT). Methods The population was defined from the Swedish Stroke Register (RS) and included all patients with an ICH diagnosis during 2016–2019 in Region Skåne. Hemorrhage volume on their initial head CT was measured using ABC/2 and manual segmentation (Sectra IDS7 volume measurement tool) and the automated volume quantification tool (qER–NCCT) by Qure.ai. The first 500 were examined by two independent readers. Results A total of 1649 ICH patients were included. The qER–NCCT had 97% sensitivity in identifying ICH. In total, there was excellent agreement between volumetric measurements of ICH volumes by qER–NCCT and manual segmentation by interclass correlation (ICC = 0.96), and good agreement (ICC = 0.86) between qER–NCCT and ABC/2 method. The qER–NCCT showed volume underestimation, mainly in large (> 30 ml) heterogenous hemorrhages. Interrater agreement by (ICC) was 0.996 (95% CI: 0.99–1.00) for manual segmentation. Conclusion Our study showed excellent agreement in volume quantification between the fully automated software qER–NCCT and manual segmentation of ICH on NCCT. The qER–NCCT would be an important additive tool by aiding in early diagnostics and prognostication for patients with ICH and in provide volumetry on a population-wide level. Further refinement of the software should address the underestimation of ICH volume seen in a portion of large, heterogenous, irregularly shaped ICHs.

Radiological features of brain hemorrhage through automated segmentation from computed tomography in stroke and traumatic brain injury

“…The most frequent locations for the ICH segmentations (in standard space) did not exceed 25% of the sample at any given site. This finding tends to align with work by others, who collectively reported on supratentorial locations, putamen, and thalamus as being common ICH locations, in addition to the cerebral hemispheres and ventricle locations ( 9 , 23 , 24 ). Future work is needed to characterize the locations of ICH in relation to radiological notes because there is renewed interest in location specific haematoma volume cut-offs in relation to outcomes ( 14 ).…”

“…The ROC analysis for the entire sample showed high performance metrics, while the performance metrics were lower for the suspected TBI site. Several other studies have reported on hemorrhage detection using CT scans ( 9 , 19 ) and these efforts used considerably larger sample sizes than the external CT cases for the current VIOLA-AI tool. It is intriguing that the VIOLA-AI tool achieved adequate classification performance for all samples, but the scores decreased when considering the low yield cases of consecutive mild TBI cases ( 29 ).…”

“…Skåne: 223 ICH cases were selected from an existing stroke database at the Skåne University Hospital in Lund/Malmö, Sweden ( 9 ) and cases were overseen/reviewed by JW and TU.…”

Lower initial red blood cell count in cerebrospinal fluid predicts good functional outcome in patients with spontaneous subarachnoid haemorrhage

“…This hypothesis is supported by our data demonstrating a correlation between CSF RBC count and both the modified Fisher score and the Hijdra ventricle score, the current state of the art to measure SAH volumes. Novel techniques to quantify intracranial blood volume are increasingly used [43][44][45], but do not necessarily improve predictive models for outcome [15]. In patients with intracerebral haemorrhage, artificial intelligence may help to rapidly quantify haematoma volume and the extension of perihaematomal oedema [46].…”

AI assisted reader evaluation in acute CT head interpretation (AI-REACT): protocol for a multireader multicase study

“… 11 In a Swedish stroke registry study, qER was found to have about 97% sensitivity in detecting non-traumatic ICH, and 95% of the false-negative ICHs were <1 mm in diameter. 28 In an unpublished study conducted for the purpose of regulatory submission, using 1320 NCCTH scans from multiple sites in the USA, the AUC was reported as over 97% for ICH, skull fracture, mass effect and midline shift. The sensitivity and specificity in detecting any one of the four abnormalities (ICH, skull fracture, mass effect, midline shift) were reported to be 98.53% (95% CI: 97.45% to 99.24%) and 91.22% (95% CI: 88.39% to 93.55%), respectively.…”