Retrospective analysis and prospective validation of an AI-based software for intracranial haemorrhage detection at a high-volume trauma centre

• DOI: 10.1038/s41598-022-24504-y
• Publication Date: 2022-11-18T20:04:46Z
• Authors: Adil Zia; Calvin Fletcher; Shalini Bigwood; Prasanna Ratnakan...; Jarrel Seah; Robin Lee; Helen Kavnoudias; Meng Law
• Keywords: Triage; Head trauma

Rapid detection of intracranial haemorrhage (ICH) is crucial for assessing patients with neurological symptoms. Prioritising these urgent scans reporting presents a challenge radiologists. Artificial intelligence (AI) offers solution to enable radiologists triage and reduce errors. This study aims evaluate the accuracy an ICH-detection AI software whether it benefits high-volume trauma centre in terms reducing diagnostic A peer review head CT performed prior implementation was conducted identify department's current miss-rate. Once implemented, validated using over one month, reviewed by neuroradiologist. The turn-around-time calculated as time taken from scan completion report finalisation. 2916 reports were part audit. flagged 20 cases that negative-by-report. Two true-misses had no follow-up imaging. Both followed up exhibited long-term sequelae. For ICH-positive scans, there increase TAT total sample (35.6%), statistically insignificant decrease emergency (- 5.1%) outpatient 14.2%) cohorts. tested on real-world data Australian centre. comparable reported literature. demonstrated institution's low miss-rate short time, therefore any improvements use would be marginal challenging measure.