Abstract
Stereotactic Arrhythmia Radiotherapy (STAR) treats ventricular tachycardia (VT) but requires internal target volume (ITV) expansions to compensate for cardiorespiratory motion. Current clinical r4DCT imaging methods are limited, and the reconstructed r4DCTs suffer from unmanaged cardiac motion artifacts that affect the quantitative assessment of respiratory motion. A groupwise surface-to-surface deformable image registration (DIR) algorithm, named gCGF, was developed. A novel principal component filtering (PCF) mechanism and a spatial smoothing mechanism were developed and incorporated into gCGF to iteratively register heart contours from an average respiratory-phase CT to ten r4DCT phases while removing random cardiac motion from the cyclic respiratory motion. Validation on digital phantoms showed that gCGF achieved a mean target registration error of 0.63±0.51 mm. Among 20 STAR patients, the heart's maximum respiratory motion magnitudes ranged from 3.6 to 7.9 mm. Significant dominance of the first principal component of the motion direction was observed.
Introduction: The Challenge of Cardiac Artifacts
Stereotactic Arrhythmia Radiotherapy (STAR) is a non-invasive treatment for ventricular tachycardia (VT) using targeted radiation to ablate arrhythmogenic heart tissue. Because the heart moves due to both breathing (respiratory) and beating (cardiac), radiation beams cannot dynamically track the target. Clinicians expand the treatment area into an Internal Target Volume (ITV) based on respiratory 4D CTs (r4DCT).
However, r4DCTs suffer from severe, unmanaged cardiac motion artifacts because the heart beats much faster than the breathing cycle. These artifacts appear as blurry, misaligned images, making it difficult to isolate respiratory motion for treatment planning. Furthermore, metal from implanted cardioverter-defibrillators (ICDs) creates additional image noise.
Methods: The gCGF Algorithm
To address these limitations, the authors developed gCGF (Groupwise Coherent Point Drift - Gaussian Mixture Model - Finite Element Model), a surface-to-surface deformable image registration (DIR) algorithm. Instead of relying on volumetric pixel intensities which are often corrupted by artifacts, gCGF registers 3D point clouds of the heart's surface.
The algorithm iteratively deforms an average heart shape to match the 10 respiratory phases of an r4DCT scan. Two key mechanisms ensure accuracy:
- Principal Component Filtering (PCF): Respiratory motion is cyclic, while cardiac artifacts in r4DCT appear as random, uncorrelated noise. By applying PCA to the Deformation Vector Fields (DVFs) and reconstructing motion using only the primary components, gCGF filters out high-frequency cardiac "jitter."
- Spatial Smoothing: To maintain biomechanical realism and surface continuity, the algorithm replaces each GMM centroid with the mean of its neighbors during update steps.
Results and Clinical Insights
The algorithm was validated using 8 digital XCAT phantoms and applied to 20 STAR patients.
- Accuracy: On phantoms, gCGF achieved a mean Target Registration Error (TRE) of 0.63 ± 0.51 mm, outperforming the state-of-the-art pTVreg method by 25%.
- Motion Quantification: For the 20 patients, the VT target's maximum respiratory motion ranged from 3.0 to 6.7 mm (mean: 0.8 to 2.9 mm). The peak-to-peak range was highly variable, spanning 4.7 to 11.8 mm.
- Directionality: Heart motion is predominantly one-dimensional, aligning with the Superior-Inferior direction driven by the diaphragm. However, significant patient-specific variability (e.g., Anterior-Posterior motion) highlights the need for personalized motion analysis rather than population-based margins.
Conclusion
The gCGF algorithm effectively isolates respiratory motion from cardiac artifacts, providing a more accurate basis for STAR treatment planning. While currently limited to surface-to-surface registration—requiring interpolation for internal targets—it represents a significant step toward safer, more precise radiation margins in cardiac ablation.
Summary updated