Brain Midline Approximation to Improve Symmetry Analysis of Brain CT Scans

Area of Science:

• Medical imaging
• Neuroscience
• Computer vision

Background:

• Brain hemisphere symmetry analysis is crucial for diagnosing neurological conditions like tumors, stroke, and Alzheimer's disease.
• Accurate determination of the brain's midline is essential for comparative hemispheric analysis in medical scans.
• Existing methods struggle with geometric deformations, artifacts, noise, and anatomical variations, necessitating robust solutions.

Purpose of the Study:

• To present a novel and robust method for estimating the brain's midline.
• To develop an algorithm that is resilient to common imaging challenges and anatomical variability.
• To provide a tool that enhances the diagnostic capabilities of medical imaging.

Main Methods:

• A two-step approach involving image reorientation and comprehensive brain volume analysis.
• Utilizing a matched ellipse to determine a head mask model and estimate skull shape.
• Employing 2D cross-correlation on edge images of successive scans to generate a precise midline.

Main Results:

• The developed algorithm achieved a 92.9% success rate in midline estimation.
• The mean error for the midline estimation was recorded at 1.2 mm.
• Results are competitive with state-of-the-art methods and demonstrate high efficiency.

Conclusions:

• The novel midline estimation method is highly effective and robust.
• Key advantages include resistance to head tilt, implementation simplicity, and high efficiency.
• The method has broad applicability in supporting medical imaging diagnostics.