HDTV: a high-order directional total variation reconstruction algorithm from sparse and limited-angle data in...

Area of Science:

• Medical Imaging
• Computational Imaging
• Biomedical Engineering

Background:

• Electron paramagnetic resonance imaging (EPRI) is crucial for adaptive radiation therapy, but long scanning times hinder its clinical application.
• Reducing EPRI scanning time is essential for widespread adoption and improved tumor control.
• Sparse reconstruction and limited-angle data acquisition are key strategies to accelerate EPRI.

Purpose of the Study:

• To develop a novel method for accelerating three-dimensional (3D) EPRI by combining sparse and limited-angle data acquisition.
• To improve the precision and imaging quality of accelerated EPRI.
• To address the limitations of existing EPRI scanning times.

Main Methods:

• Proposed a high-order directional total variation (HDTV) model with high-order constraints to suppress staircase artifacts.
• Derived the Chambolle-Pock (CP) algorithm for solving the HDTV model.
• Implemented sparse acquisition using limited-angle data for accelerated 3D EPRI.

Main Results:

• The HDTV-CP algorithm effectively suppresses artifacts from limited-angle and sparse data, preserving image details.
• Validated on simulation and real data, demonstrating superior performance over classic total variation (TV) methods.
• Achieved a significant reduction in normalized root mean square error (nRMSE) from 0.34 to 0.16 and an increase in Pearson correlation coefficient (PCC) from 0.93 to 0.98.

Conclusions:

• Successfully combined limited-angle and sparse reconstruction challenges in EPRI for the first time.
• The HDTV method offers potential for up to 16x acceleration in EPRI imaging without compromising quality in specific scenarios.
• The developed techniques have potential applications in limited-angle computed tomography (CT) image reconstruction.