Validating Artificial Intelligence Model to Optimize Detection of Intracerebral Hemorrhage

Introduction: Artificial intelligence (AI) can alert the radiologist to the presence of ischemic stroke as fast as 1 to 2 minutes from scan completion, leading to faster diagnosis and treatment. Thus, we aimed to validate a new AI model called Viz.ai ICH to improve the detection of intracerebral hemorrhage (ICH).

Method(s): We performed a retrospective database analysis of 4,203 consecutive noncontrast brain CT reports between September to December 2021 within a single institution. The reports were made by neuroradiologists who reviewed each case for ICH. Each positive case was categorized based on subtype, timing, and size/volume. The AI model was validated by assessing its diagnostic performance with Viz.ai ICH as the index test compared with the neuroradiologists’ interpretation as the gold standard.

Result(s): 387 of 4,203 noncontrast brain CT reports were positive for ICH. The overall sensitivity of Viz.ai ICH was 68%, specificity was 99%, PPV was 90%, and NPV was 97%. Subgroup analysis revealed sensitivity improves with higher acuity and volume/size across ICH subtypes.

Conclusion(s): Our analysis indicates that AI can accurately detect the presence of ICH particularly for large volume/size ICH. With improvements in the AI algorithm, radiologists can detect ICH more effectively.

Publication History

Article published online:
09 February 2023