Related Experiment Video
Updated: Jun 28, 2026
14:08
Automated Midline Shift and Intracranial Pressure Estimation based on Brain CT Images
Published on: April 13, 2013
A lightweight improved YOLOv11 framework for intracranial hemorrhage detection
Yunchang Zheng1, Mingzhe Du1, Qing Chang1
1College of Electrical Engineering, Hebei University of Architecture, Zhangjiakou, China.
Scientific Reports
|June 26, 2026
Summary
This study introduces a lightweight YOLOv11-based framework for detecting intracranial hemorrhage in CT scans. The enhanced model improves accuracy and efficiency for real-time medical diagnosis.
Area of Science:
- Medical Imaging
- Artificial Intelligence
- Computer Vision
Background:
- Accurate intracranial hemorrhage detection in CT images is critical for prompt treatment and improved patient outcomes.
- Current deep learning models face challenges with subtle lesions, multi-scale features, class imbalance, and background noise in CT scans.
- Real-time detection is hindered by the trade-off between precision and computational efficiency in existing methods.
Purpose of the Study:
- To develop a lightweight and efficient deep learning framework for improved intracranial hemorrhage detection in CT images.
- To enhance feature extraction, fusion, and localization capabilities for better lesion identification.
- To address limitations in existing models regarding subtle lesion characteristics and computational efficiency.
Main Methods:
- A modified YOLOv11 framework incorporating RepStem for enhanced early feature representation.
- Introduction of the FDPN-DASI module for improved multi-scale hematoma aggregation and background suppression.
- Replacement of C2PSA blocks with cascaded C2CGA attention to better detect small hemorrhages and ambiguous boundaries.
Main Results:
- The proposed method demonstrated superior performance compared to the YOLOv11 baseline on the BHX dataset.
- Achieved significant improvements in precision (3.9%) and recall (3.4%) over the baseline.
- Increased mean Average Precision (mAP@0.5) by 3.7% and mAP@0.5:0.95 by 4.6% while maintaining a lightweight model.
Conclusions:
- The developed lightweight YOLOv11-based framework offers enhanced accuracy and efficiency for intracranial hemorrhage detection.
- The model's lightweight nature (2.65M parameters) and high inference speed (864.86 FPS) make it suitable for real-time CT-based computer-aided diagnosis.
- This approach shows significant potential for improving emergency intervention and reducing patient mortality and disability.
Related Concept Videos
01:17Hemorrhagic Stroke l: Introduction
A hemorrhagic stroke is an acute neurological event that occurs when a weakened cerebral blood vessel ruptures, allowing blood to accumulate within or around the brain. The sudden release of blood forms a focal hematoma that increases intracranial pressure, displaces neural tissue, and can obstruct cerebrospinal fluid pathways. These effects may be compounded by intraventricular extension of the hemorrhage, cerebral edema, or compression of adjacent structures, all of which contribute to...
01:19Cerebral Edema l: Introduction
Cerebral edema is a pathological increase in brain water content that disrupts intracranial pressure regulation and impairs neurological function. Because the cranial vault is rigid, even modest increases in tissue volume can compromise cerebral perfusion, distort neural structures, and initiate secondary injury. Cerebral edema develops through four principal mechanisms: vasogenic, cytotoxic, interstitial, and ionic.Vasogenic EdemaVasogenic edema arises from disruption of the blood–brain...
01:29Hemorrhagic Stroke ll: Pathophysiology
A hemorrhagic stroke develops when a cerebral blood vessel ruptures, allowing blood to escape into the surrounding brain tissue, as in intracerebral hemorrhage (ICH), or into the subarachnoid space, as in subarachnoid hemorrhage (SAH). Because the skull is a rigid compartment, the sudden presence of extravascular blood rapidly increases intracranial pressure and compresses adjacent neural structures, leading to immediate tissue injury and impaired cerebral perfusion.Mass Effect and Primary...
01:14Increased Intracranial Pressure l: Introduction
Intracranial hypertension is a sustained elevation of intracranial pressure (ICP) above 22 mm Hg. In supine adults, normal ICP is ~7–15 mm Hg.The rigid, nonexpandable cranium contains three components—brain tissue, blood, and cerebrospinal fluid (CSF)—that total ~1,700 mL in a typical adult: 1,400 mL brain (~80%), 150 mL blood (~10%), and 150 mL CSF (~10%). According to the Monro–Kellie doctrine, total intracranial volume is effectively fixed. When one component expands, CSF and venous blood...
01:19Imaging Studies VII: Vascular Imaging
DefinitionRenal angiography, also known as renal arteriography, is an imaging technique used to obtain a comprehensive view of blood flow and the vascular structure of blood vessels in the kidneys and surrounding areas.PurposeRenal angiography detects blood vessel abnormalities in the kidneys, such as aneurysms, stenosis, thrombosis, vascular tumors, and renal artery stenosis. It evaluates kidney function and guides interventional treatments like angioplasty or stent placement.Pre-Procedure...