Online tree-structure-constrained RPCA for background subtraction of X-ray coronary angiography images

Area of Science:

• Medical Imaging
• Image Processing
• Cardiovascular Imaging

Background:

• X-ray coronary angiography (XCA) is crucial for diagnosing coronary vessel diseases.
• Complex and dynamic backgrounds in XCA hinder accurate diagnosis.
• Existing background subtraction methods face challenges with intricate XCA data.

Purpose of the Study:

• To develop an advanced method for effective XCA background subtraction.
• To improve the clarity and diagnostic value of XCA images.
• To address the challenges posed by complex and dynamic XCA backgrounds.

Main Methods:

• Proposed an online tree-structure-constrained robust PCA (OTS-RPCA) method.
• Utilized morphological closing for pre-processing to remove large structures.
• Decomposed XCA sequences into low-rank background, residual dynamic background, and vascular foreground subspaces.
• Introduced a tree-structured norm for vessel spatial coherency and adaptive regularization for vessel area changes.

Main Results:

• Evaluated on real clinical and synthetic low-contrast XCA datasets.
• Achieved high performance metrics: global CNR (6.27), local CNR (3.07), and SSIM (0.97) on real data.
• Demonstrated superiority over seven state-of-the-art methods in enhancing coronary vessel contrast and structure preservation.

Conclusions:

• The OTS-RPCA method accurately subtracts backgrounds from XCA images.
• This technique can potentially reduce contrast agent dose and the number of injections in coronary interventions.
• The method offers a promising basis for improved coronary interventions.