سامانه همایش‌ها TSTA | ثبت‌نام و ارسال مقاله کنفرانس‌ها و کنگره‌های ملی

صفحه اصلی / سی و دومین کنفرانس ملی و دهمین کنفرانس بین المللی مهندسی زیست پزشکی ایران

Geometry-Aware Anisotropic Total Variation Regularization for Limited-View Photoacoustic Tomography

نویسندگان :

Amirreza Jodeiry¹ Zahra Kavehvash²

1- دانشگاه صنعتی شریف، تهران، ایران 2- دانشگاه صنعتی شریف، تهران، ایران

کلمات کلیدی :

Photoacoustic Tomography،Limited-View،Model-Based Reconstruction،Artifact Suppression

چکیده :

Photoacoustic tomography (PAT) is a hybrid imaging modality that combines optical absorption contrast with ultrasonic detection, providing high-resolution visualization of tissue structures. However, PAT reconstructions often suffer from artifacts under limited-view acquisition due to incomplete angular sampling. Conventional isotropic total variation (TV) applies uniform smoothing, which can blur structural details, while image-adaptive anisotropic methods overlook the measurement geometry, potentially amplifying artifacts along poorly sampled orientations. We propose a geometry-aware anisotropic total variation (GA-TV) regularization that incorporates the angular sensitivity of the sensor array into the reconstruction process. GA-TV computes directional derivatives and applies geometry-derived weights to selectively suppress gradients in under-sampled directions while preserving edges in well-sampled orientations. The method integrates seamlessly with model-based (MB) reconstruction frameworks. Simulations with vessel and brain phantoms using full-ring, half-ring, and linear arrays show that GA-TV improves artifact suppression and structural fidelity. Quantitative evaluation using SSIM and ASR confirms superior reconstruction quality, achieving a 27% improvement in ASR and higher SSIM compared to conventional methods.