黑料福利网

Adenomyosis is particularly difficult to diagnose because the abnormal tissue tends to blend into the surrounding healthy muscle, making it hard to detect using standard imaging techniques like B-mode ultrasound or MRI. To improve diagnostic accuracy, the Morphological Uterus Sonographic Assessment (MUSA) criteria were developed. These criteria help identify features such as myometrial thickening, cysts, and irregular vascular patterns, even when the tissue lacks clear boundaries.

CEUS and CUDI: a new frontier in imaging

Given the role of angiogenesis (formation of new blood vessels) in adenomyosis, Contrast-Enhanced Ultrasound (CEUS) has emerged as a promising imaging technique. CEUS uses microbubbles that remain in the bloodstream and reflect ultrasound waves in a unique way, allowing for real-time visualization of blood flow and vascular structures.

At 黑料福利网, researchers have developed the Contrast-Ultrasound Dispersion Imaging (CUDI) framework to analyze how these microbubbles disperse through the uterine vasculature. CUDI models this dispersion as a convective process through porous tissue, enabling the quantification of vascular changes. Two key approaches are used:

• Time-Intensity Curve (TIC) fitting: Tracks contrast agent concentration over time at each pixel.
• Spatiotemporal similarity analysis: Uses metrics like temporal correlation and spectral coherence to map dispersion patterns.

The problem of motion: why it matters

Despite the advantages of CEUS and CUDI, uterine motion during imaging poses a major challenge. Movements caused by natural uterine contractions, bowel activity, or bladder filling can distort the images, leading to blurred vascular patterns and unreliable data. These motion artifacts reduce the accuracy of TIC fitting and compromise the diagnostic value of CEUS.

Even with external stabilization methods like probe holders, internal body motion remains a significant issue. This thesis addresses the need for motion compensation to improve the consistency and quality of CEUS imaging.

Motion compensation: a two-stage solution

To tackle motion artifacts, a two-stage model-based motion correction framework was developed:

1. Rigid Correction: Aligns the entire image to correct for global shifts.
2. Non-Rigid Correction: Adjusts for local deformations within the tissue.

The effectiveness of these corrections was evaluated using metrics such as goodness-of-fit (r虏), Normalized Cross-Correlation (NCC), and Weighted Temporal Cross-Correlation (WTCC). The impact of down-sampling on correction performance was also studied.

Key findings and future directions

• Rigid correction improved TIC fitting, especially in the far-field region, with better results when down-sampling was avoided.
• Non-rigid correction, however, reduced the number of usable fits, indicating that this method needs further refinement.
• Singular Value Decomposition (SVD) was identified as a time-efficient alternative to full TIC fitting, offering reasonable motion correction with less computational effort.

These findings suggest that optimizing motion correction strategies can significantly enhance the diagnostic power of CEUS and CUDI, especially for complex conditions like adenomyosis.

Program: Electrical Engineering

Research Group: Biomedical Diagnostics Lab, Signal Processing Systems