Bi-exponential magnetic resonance signal model for partial volume computation

Area of Science:

• Medical Imaging
• Biophysics
• Computational Neuroscience

Background:

• Partial volume (PV) effects in magnetic resonance (MR) imaging limit accurate quantification of small brain structures.
• These effects arise when a single voxel contains multiple tissue types, complicating analysis of structures like gray matter (GM) and white matter (WM).
• Existing methods struggle with subvoxel accuracy, necessitating improved techniques for precise anatomical measurements.

Purpose of the Study:

• To develop and validate a novel method for computing partial volume (PV) effects in MR images with enhanced subvoxel accuracy.
• To model MR signal using a biexponential linear combination to represent tissue contributions within each voxel.
• To assess the performance of the proposed method against traditional approaches using physical phantoms and simulated brain data.

Main Methods:

• MR signal modeling using a biexponential linear combination to account for up to two tissue types per voxel.
• Estimation of tissue-specific parameters (T1, T2, proton density).
• Solving a linear system to retrieve fractional tissue contents (magnetizations) for PV calculation.

Main Results:

• Preliminary validation on a custom physical phantom designed for PV effect studies.
• Testing on BrainWeb simulated brain images to estimate GM and WM PV effects.
• The proposed method demonstrated superior performance, outperforming traditional methods by 33% for GM and 34% for WM in terms of root mean squared error.

Conclusions:

• The biexponential modeling approach offers a significant improvement in accuracy for partial volume quantification in MR imaging.
• This method provides a more precise way to analyze brain anatomy, particularly for structures affected by PV effects.
• The enhanced accuracy has implications for various neuroimaging applications requiring detailed structural quantification.