Related Experiment Video
Updated: Jul 5, 2026
Automated Midline Shift and Intracranial Pressure Estimation based on Brain CT Images
Published on: April 13, 2013
A deep learning algorithm for automatic detection and classification of acute intracranial hemorrhages in head CT
Xiyue Wang1, Tao Shen2, Sen Yang2
1College of Computer Science, Sichuan University, Chengdu 610065, China.
Insights
Artificial intelligence aids in diagnosing acute intracranial hemorrhage (ICH) on CT scans. A deep learning model achieved expert-level accuracy, winning a major competition and showing strong generalization for clinical use.
Area of Science:
- Medical Imaging
- Artificial Intelligence
- Radiology
Background:
- Acute intracranial hemorrhage (ICH) is a critical condition requiring prompt diagnosis via non-contrast head CT.
- Radiologist accuracy in detecting ICH on CT can be variable due to subtle findings and workload pressures.
- Artificial intelligence (AI) offers potential to enhance diagnostic speed and accuracy in medical imaging.
Purpose of the Study:
- To develop and evaluate a deep learning algorithm for accurate detection and subtyping of acute ICH on CT scans.
- To create an AI tool that mimics radiologist interpretation for improved decision-making.
- To assess the performance of the AI model against expert radiologists and on external datasets.
Main Methods:
- A deep learning approach combining a 2D CNN and two sequence models was designed.
- The model was trained on the large-scale 2019-RSNA Brain CT Hemorrhage Challenge dataset (>25,000 scans).
- Performance was evaluated using Area Under the Curve (AUC) metrics and validated on independent datasets.
Main Results:
- The AI model achieved high AUCs for ICH detection (0.988) and subtype classification (e.g., IVH 0.996, IPH 0.992).
- The method secured 1st place in the RSNA challenge against 1345 international teams.
- Excellent performance was maintained on external validation datasets (AUCs 0.964 and 0.949).
Conclusions:
- The developed deep learning algorithm demonstrates expert-level accuracy and robust generalization for acute ICH detection and subtyping.
- The AI tool shows significant potential as a second-read or triage system in clinical settings.
- This AI approach can assist radiologists, leading to more prompt and accurate diagnoses of acute intracranial hemorrhage.
Abstract:
Acute Intracranial hemorrhage (ICH) is a life-threatening disease that requires emergency medical attention, which is routinely diagnosed using non-contrast head CT imaging. The diagnostic accuracy of acute ICH on CT varies greatly among radiologists due to the difficulty of interpreting subtle findings and the time pressure associated with the ever-increasing workload. The use of artificial intelligence technology may help automate the process and assist radiologists for more prompt and better decision-making. In this work, we design a deep learning approach that mimics the interpretation process of radiologists, and combines a 2D CNN model and two sequence models to achieve accurate acute ICH detection and subtype classification. Being developed using the extensive 2019-RSNA Brain CT Hemorrhage Challenge dataset with over 25000 CT scans, our deep learning algorithm can accurately classify the acute ICH and its five subtypes with AUCs of 0.988 (ICH), 0.984 (EDH), 0.992 (IPH), 0.996 (IVH), 0.985 (SAH), and 0.983 (SDH), respectively, reaching the accuracy level of expert radiologists. Our method won 1st place among 1345 teams from 75 countries in the RSNA challenge. We have further evaluated our algorithm on two independent external validation datasets with 75 and 491 CT scans, respectively, and our method maintained high AUCs of 0.964 and 0.949 for acute ICH detection. These results have demonstrated the high performance and robust generalization ability of our proposed method, which makes it a useful second-read or triage tool that can facilitate routine clinical applications.
