Related Experiment Videos

Segmentation of MRS signals using ASPECT (analysis of SPectra using Eigenvector Decomposition of Targets)

J R Roebuck1, J P Windham, D O Hearshen

  • 1Bioengineering Program, University of Michigan, Ann Arbor 48109.

Medical Physics
|February 1, 1994
PubMed

Insights

This study introduces a linear model to segment in vivo magnetic resonance spectroscopy (MRS) signals affected by partial volume contamination (PVC). The model estimates partial volume contributions and spatially selective signals, improving data analysis in the human brain.

Area of Science:

  • Magnetic Resonance Spectroscopy (MRS)
  • Medical Imaging Analysis
  • Biomedical Signal Processing

Background:

  • Partial volume contamination (PVC) significantly impacts in vivo MRS data quality.
  • Current methods focus on improving spatial resolution, but PVC affects spectra regardless of resolution.

Purpose of the Study:

  • To introduce a novel linear model for segmenting in vivo MRS signals compromised by PVC.
  • To estimate partial volume contributions and spatially selective signal components within voxels.

Main Methods:

  • Developed a linear segmentation algorithm based on image processing filters.
  • Utilized first principles for model development and computer simulations for evaluation.
  • Applied the model to multivoxel or chemical shift imaging data at any spatial resolution.

Main Results:

  • The model effectively estimates the size of partial volumes contributing to PVC-affected voxels.
  • It can determine spatially selective signal components expected from these partial volumes.
  • Identified spectral perturbants that violate linearity assumptions and cause systematic errors.

Conclusions:

  • The developed model provides a method for segmenting in vivo MRS signals affected by PVC.
  • It is applicable to various data types, including 1H-CSI from the human brain.
  • Understanding and accounting for model limitations is crucial for accurate in vivo applications.

Related Concept Videos