Observations on maximum entropy processing of MR images

Area of Science:

• Medical Imaging
• Image Processing
• Information Theory

Background:

• Previous Maximum Entropy (MAXENT) criteria demonstrated suboptimal performance in Magnetic Resonance (MR) image processing.
• Existing MAXENT formulations intermixed two distinct data transmission applications: pixel-location and pixel-value data.
• The distinction between transmitted variables (pixel-locations vs. pixel-values) was not adequately addressed in prior literature.

Purpose of the Study:

• To differentiate between the two "data transmission" applications within MR image processing.
• To modify MAXENT criteria by focusing on pixel-value information entropy for improved MR image processing.
• To demonstrate the successful application of modified MAXENT criteria in restoring MR image visualization.

Main Methods:

• Distinguished between pixel-location entropy and pixel-value information entropy in MR image data.
• Modified MAXENT criteria to specifically utilize pixel-value intensities for data processing.
• Applied the modified MAXENT criteria to restore display visualization from nonuniform surface coils in lumbar spine scans.

Main Results:

• The modified MAXENT criteria, utilizing pixel-value information entropy, yielded successful data processing results.
• Effective restoration of display visualization was achieved for lumbar spine scans acquired with nonuniform surface coils.
• The study confirmed the enhanced utility of MAXENT in specific MR image processing contexts.

Conclusions:

• Revising MAXENT criteria to account for pixel-value information entropy significantly improves MR image processing performance.
• The distinction between data transmission types is crucial for successful application of MAXENT in medical imaging.
• Modified MAXENT provides a valuable tool for enhancing visualization in challenging MR imaging scenarios, exemplified by lumbar spine scans.