Multi-granularity Adversarial Generation Integrated Consistency Representation for Chest Low-Contrast-Enhanced CT...

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Computer Vision

Background:

• Contrast-enhanced computed tomography (CECT) is crucial for assessing vascular structures and function.
• High contrast agent doses pose risks of renal injury, while low doses compromise diagnostic image quality.
• Existing CECT synthesis methods struggle with edge clarity, contrast anomalies, and texture distortion.

Purpose of the Study:

• To develop a novel method, MAGIC (Multi-granularity Adversarial Generation Integrated Consistency Representation), for synthesizing high-quality CECT from low-contrast enhanced CT.
• To address limitations of existing methods, including edge unclarity, contrast anomalies, and texture distortion.

Main Methods:

• Proposed MAGIC, incorporating a Multi-Granularity Refined Booster (MRB) for enhanced feature representation and edge clarity.
• Introduced a Supervised Contrast Enhancement Module (SCEM) to adaptively adjust contrast and overcome anomalies.
• Implemented Hierarchical Harmonized Consistency Representation (HHCR) for improved texture restoration using domain consistency.
• Utilized a Dual-path Dynamic Collaborative Discriminator (DDCD) for comprehensive fidelity evaluation.

Main Results:

• MAGIC demonstrated superior performance in edge clarity, image contrast, and texture restoration compared to existing methods.
• Qualitative and quantitative analyses confirmed the effectiveness of the proposed innovations.
• The method shows significant potential for clinical application in CECT synthesis.

Conclusions:

• MAGIC offers a robust solution for generating high-quality CECT from low-dose scans, mitigating risks associated with high contrast agent doses.
• The method effectively enhances image clarity, contrast, and texture, addressing key limitations of prior approaches.
• MAGIC holds substantial promise for improving the safety and diagnostic accuracy of CECT examinations.