Related Experiment Video
Updated: Sep 27, 2026
Automated Midline Shift and Intracranial Pressure Estimation based on Brain CT Images
Published on: April 13, 2013
Feasible Study on Intracranial Hemorrhage Detection and Classification using a CNN-LSTM Network
Insights
This study introduces a deep learning method for automatically identifying and classifying intracranial hemorrhage (ICH) subtypes from head CT scans. The AI model achieved high accuracy, aiding radiologists in diagnosing brain bleeds.
Area of Science:
- Medical Imaging
- Artificial Intelligence
- Radiology
Background:
- Intracranial hemorrhage (ICH) is a critical condition with diverse causes, including stroke, trauma, and hypertension.
- Accurate and timely identification of ICH subtypes is crucial for patient outcomes.
- Current diagnostic methods rely on manual interpretation of head CT scans, which can be time-consuming.
Purpose of the Study:
- To develop and assess a deep learning model for the automatic identification and classification of intracranial hemorrhage (ICH) subtypes.
- To evaluate the feasibility of using head CT images for AI-driven ICH diagnosis.
- To enhance the accuracy and efficiency of ICH detection and classification in clinical practice.
Main Methods:
- A deep learning approach utilizing a CNN-LSTM model was employed for ICH classification.
- The model was trained on 4,516,842 head CT images after applying windowing techniques (brain, bone, subdural).
- The Xception model served as the deep CNN backbone, with LSTM configured with 64 nodes and 32 timesteps.
Main Results:
- The model was tested on 727,392 head CT images.
- The automated system achieved a weighted multi-label logarithmic loss of 0.07528.
- The proposed method demonstrated high accuracy in identifying and classifying ICH subtypes.
Conclusions:
- The developed deep learning technique shows significant potential for the automatic identification and classification of ICH.
- This AI-driven approach can assist radiologists in interpreting head CT scans, improving diagnostic accuracy.
- The method offers a valuable tool for quantitative analysis in neuroimaging and brain-related conditions.
Abstract:
Intracranial hemorrhage (ICH) is a life-threatening condition, the outcome of which is associated with stroke, trauma, aneurysm, vascular malformations, high blood pressure, illicit drugs and blood clotting disorders. In this study, we presented the feasibility of the automatic identification and classification of ICH using a head CT image based on deep learning technique. The subtypes of ICH for the classification was intraparenchymal, intraventricular, subarachnoid, subdural and epidural. We first performed windowing to provide three different images: brain window, bone window and subdural window, and trained 4,516,842 head CT images using CNN-LSTM model. We used the Xception model for the deep CNN, and 64 nodes and 32 timesteps for LSTM. For the performance evaluation, we tested 727,392 head CT images, and found the resultant weighted multi-label logarithmic loss was 0.07528. We believe that our proposed method enhances the accuracy of ICH identification and classification and can assist radiologists in the interpretation of head CT images, particularly for brain-related quantitative analysis.
