Related Experiment Videos
Interplay between intensity standardization and inhomogeneity correction in MR image processing.
Anant Madabhushi1, Jayaram K Udupa
1Department of Biomedical Engineering, Rutgers University, 617 Bowser Road, Rm. 102, BME Bldg., Piscataway, NJ 08854, USA. anantm@rci.rutgers.edu
IEEE Transactions on Medical Imaging
|May 14, 2005
Summary
In magnetic resonance (MR) image analysis, applying inhomogeneity correction before intensity standardization significantly improves image quality. This sequence enhances standardization without negatively impacting correction, offering optimal results for MR images.
Area of Science:
- Medical Imaging
- Image Processing
- Magnetic Resonance Imaging (MRI)
Background:
- Image intensity standardization corrects signal variations in MR images.
- Inhomogeneity correction addresses low-frequency background nonuniformities in MR images.
- The combined effects and optimal sequence of these postprocessing steps remain understudied.
Purpose of the Study:
- To evaluate the impact of applying inhomogeneity correction and intensity standardization in different sequences on MR image quality.
- To determine the optimal order of these postprocessing techniques for enhancing MR image analysis.
- To investigate the interdependency and potential biases between these two correction methods.
Main Methods:
- Experiments were conducted on clinical and phantom MR datasets (nearly 4000 3D images) across four MRI protocols.
- Artificial nonstandardness and inhomogeneity levels were introduced to systematically assess correction performance.
- The study compared the sequence of inhomogeneity correction followed by standardization versus standardization followed by correction.
Main Results:
- Preceding standardization with inhomogeneity correction leads to improved standardization outcomes.
- Inhomogeneity correction biases the effect of standardization, a bias independent of the correction method used.
- Standardization does not significantly influence the inhomogeneity correction process; applying them in either order yields similar correction results.
- Repeated application of correction and standardization sequences does not substantially enhance image quality.
Conclusions:
- The optimal sequence for enhancing MR image quality and computational efficiency is inhomogeneity correction followed by intensity standardization.
- This recommended sequence ensures improved standardization and mitigates biases introduced by the correction process.
- The findings are robust across various datasets, protocols, and artificial artifact levels, validating the proposed workflow for MR image postprocessing.