Related Experiment Video
Updated: Jun 19, 2026
14:08
Automated Midline Shift and Intracranial Pressure Estimation based on Brain CT Images
Published on: April 13, 2013
Automated ventricular systems segmentation in brain CT images by combining low-level segmentation and high-level
Wenan Chen1, Rebecca Smith, Soo-Yeon Ji
1Department of Computer Science, Virginia Commonwealth University, Richmond, VA, USA. chenw6@vcu.edu
BMC Medical Informatics and Decision Making
|November 7, 2009
Summary
This study presents an automated method for segmenting brain ventricles in CT scans, crucial for diagnosing Traumatic Brain Injuries (TBI). The reliable algorithm accurately identifies ventricles, aiding faster TBI diagnosis and treatment.
Area of Science:
- Medical Imaging
- Computational Anatomy
- Radiology
Background:
- Accurate analysis of CT brain scans is critical for diagnosing and treating Traumatic Brain Injuries (TBI).
- Automated processing of CT scans can expedite diagnosis, reduce healthcare costs, and minimize human error.
- Segmentation of brain ventricles provides quantitative diagnostic information.
Purpose of the Study:
- To develop and evaluate an automated method for segmenting and identifying ventricular systems in CT brain images.
- To improve the accuracy and efficiency of Traumatic Brain Injury (TBI) diagnosis through enhanced image analysis.
Main Methods:
- CT slices alignment using skull symmetry and anatomical features for ideal midline detection.
- A two-step ventricle segmentation approach: low-level pixel segmentation (Iterated Conditional Mode and Maximum A Posteriori Spatial Probability) followed by template matching.
- Validation using a large dataset of mild and severe TBI cases.
Main Results:
- Achieved over 95% accuracy in ideal midline detection.
- 100% sensitivity in ventricle identification across all slices.
- A false positive rate of 8.59% for ventricle recognition.
- Comparison of Iterated Conditional Mode (ICM) and Maximum A Posteriori Spatial Probability (MASP) algorithms.
Conclusions:
- The proposed algorithms demonstrate reliability for automated CT brain scan analysis.
- Novelty lies in using anatomical features for midline detection and a two-step segmentation process.
- Improvements include accurate midline detection and precise ventricle recognition using anatomical features and MRI-derived templates.
