Related Experiment Video
Updated: Aug 3, 2025
09:14
Pre-Chiasmatic, Single Injection of Autologous Blood to Induce Experimental Subarachnoid Hemorrhage in a Rat Model
Published on: June 18, 2021
2.4K
Deep learning based automatic detection algorithm for acute intracranial haemorrhage: a pivotal randomized clinical
Tae Jin Yun1,2, Jin Wook Choi3, Miran Han4
1Institute of Radiation Medicine, Seoul National University Medical Research Center, Seoul, Republic of Korea.
NPJ Digital Medicine
|April 7, 2023
Summary
An artificial intelligence (AI) algorithm significantly improves the diagnosis of acute intracranial haemorrhage (AIH) on brain CT scans. AI assistance enhanced diagnostic accuracy, especially for non-radiologist physicians.
Area of Science:
- Medical Imaging
- Artificial Intelligence in Medicine
- Neurology
Background:
- Acute intracranial haemorrhage (AIH) is a critical medical emergency requiring rapid diagnosis.
- Accurate interpretation of brain computed tomography (CT) images is vital for timely management.
- Existing diagnostic workflows can be time-consuming and may benefit from technological assistance.
Purpose of the Study:
- To develop and validate an artificial intelligence (AI) algorithm for diagnosing AIH using brain CT images.
- To assess the impact of AI assistance on the diagnostic performance of various medical professionals.
Main Methods:
- A retrospective, multi-reader, pivotal, crossover, randomised study design was employed.
- An AI algorithm was trained on a large dataset (104,666 slices, 3010 patients).
- Nine reviewers (non-radiologists, radiologists, neuroradiologists) evaluated 12,663 brain CT slices with and without AI assistance.
Main Results:
- AI-assisted interpretation significantly improved overall diagnostic accuracy (0.9703 vs. 0.9471, p < 0.0001).
- Non-radiologist physicians showed the most substantial improvement in diagnostic accuracy with AI assistance.
- Board-certified radiologists also demonstrated significantly higher accuracy with AI assistance; neuroradiologists showed a non-significant trend.
Conclusions:
- AI-assisted interpretation of brain CT scans enhances diagnostic accuracy for acute intracranial haemorrhage.
- The AI tool provides the most significant benefit to non-radiologist physicians, potentially broadening access to expert-level interpretation.
- AI algorithms show promise in improving efficiency and accuracy in emergency neuroradiological diagnostics.
