Compensated Diffusion Magnetic Resonance Imaging

Edward Li, Mohammad Shafiee, Ameneh Boroomand, Farzad Khalvati, Masoom A. Haider, Alexander WongVision 2015 · DOI

Abstract

<p>This paper presents a novel compensated diffusion magnetic resonance imaging (cdMRI) system for improved tissue detail and contrast for screening and diagnosis. The proposed cdMRI system incorporates the intrinsic properties of the MRI apparatus to compensate for artifacts and degradation caused through the imaging process. Experimental results for prostate imaging show that significant improvements in tissue detail and contrast can be obtained compared to current MRI systems.</p>

Introduction

Diffusion MRI (dMRI) is a powerful non-invasive tool used to map the diffusion of water molecules in biological tissue, which is critical for identifying cancerous regions. However, the physical apparatus of the MRI machine—its magnets, coils, and electronics—introduces intrinsic degradations and artifacts. These hardware limitations result in blurred tissue details and reduced contrast, which can hinder a radiologist's ability to make an accurate diagnosis. While many methods focus on post-processing or simple denoising, this paper addresses the root cause: the mismatch between the ideal signal and what the hardware actually captures.

Methodology

The authors propose a two-stage approach to "compensate" for these hardware limitations by integrating machine-specific calibration into a probabilistic reconstruction framework.

Calibration and Compensation

Before imaging, a calibration process is performed to characterize the specific intrinsic properties of the MRI machine. This results in a compensation function ($C$) that describes how the machine alters the signal, effectively creating a mathematical "antidote" to the hardware's degradation function ($F$).

CD-SFCRF Reconstruction

The core of the paper is an extended Cross-Domain Stochastically Fully-Connected Random Field (CD-SFCRF) model. Reconstruction is treated as a Maximum a Posteriori (MAP) problem, where the goal is to find the most likely image ($Y$) given the raw $k$-space measurements ($X$).

The CD-SFCRF model architecture, bridging the frequency (k-space) and spatial (image) domains to ensure physical consistency.

The model uses two types of "potentials" to guide the reconstruction:

  • Unary Potentials: These ensure the reconstructed image stays true to the raw data while incorporating the compensation function ($C$) to counteract the machine's degradation.
  • Pairwise Potentials: These enforce spatial and data consistency, ensuring that neighboring pixels are coherent and preventing the reconstruction from becoming noisy.

Results and Discussion

The cdMRI system was evaluated using prostate imaging data, comparing standard dMRI reconstructions against the new hardware-aware approach. The results demonstrate that by explicitly modeling the machine's flaws, it is possible to recover lost detail that standard algorithms ignore.

Qualitative comparison between standard dMRI (left) and cdMRI (right) in prostate imaging, showing enhanced contrast and sharper tissue boundaries.

As seen in the visual comparisons, cdMRI images show much sharper boundaries and internal tissue structures within the prostate compared to the relatively "muddy" standard dMRI images. This significant increase in contrast is vital for providing radiologists with the information needed for clinical decision support.

Conclusion and Future Work

This paper demonstrates that a hardware-aware reconstruction framework can significantly improve the diagnostic quality of diffusion MRI. While the current results are primarily qualitative, the visual improvements in tissue detail are striking. The authors plan to adapt the cdMRI methodology to other MRI modalities, including Correlated Diffusion Imaging (CDI) and T1/T2-weighted imaging, suggesting this approach could serve as a universal upgrade for clinical MRI systems.

Summary updated

Abstract

&lt;p&gt;This paper presents a novel compensated diffusion magnetic resonance&lt;br /&gt;imaging (cdMRI) system for improved tissue detail and contrast&lt;br /&gt;for screening and diagnosis. The proposed cdMRI system&lt;br /&gt;incorporates the intrinsic properties of the MRI apparatus to compensate&lt;br /&gt;for artifacts and degradation caused through the imaging&lt;br /&gt;process. Experimental results for prostate imaging show that significant&lt;br /&gt;improvements in tissue detail and contrast can be obtained&lt;br /&gt;compared to current MRI systems.&lt;/p&gt;
BibTeX
@inproceedings{li2015compensated,
  author = {Edward Li and Mohammad Shafiee and Ameneh Boroomand and Farzad Khalvati and Masoom A. Haider and Alexander Wong},
  title = {{Compensated Diffusion Magnetic Resonance Imaging}},
  booktitle = {Vision},
  year = {2015},
  doi = {10.15353/vsnl.v1i1.47},
}
Figure