ASBA: A-line State Space Model and B-line Attention for Sparse Optical Doppler Tomography Reconstruction

TL;DR

ASBA tackles sparse ODT reconstruction by introducing an A-line ROI State Space Model (A-RS) and a B-line Phase Attention (B-PA) mechanism within a two-branch magnitude-phase network. It uses ROI-guided masking to focus flow-related information along depth and a phase-difference–driven attention along B-lines, trained with a flow-aware weighted loss to prioritize flow reconstruction. On awake and anesthetized mouse cortex datasets, ASBA consistently surpasses state-of-the-art sparse reconstruction methods in PSNR and SSIM, especially preserving small-vessel flows, while maintaining lower computational cost. This approach reduces scanning time and storage needs, enabling dynamic flow analysis in Optical Doppler Tomography with high sparsity levels such as $ imes 8$ and $ imes 16$.

Abstract

Optical Doppler Tomography (ODT) is an emerging blood flow analysis technique. A 2D ODT image (B-scan) is generated by sequentially acquiring 1D depth-resolved raw A-scans (A-line) along the lateral axis (B-line), followed by Doppler phase-subtraction analysis. To ensure high-fidelity B-scan images, current practices rely on dense sampling, which prolongs scanning time, increases storage demands, and limits the capture of rapid blood flow dynamics. Recent studies have explored sparse sampling of raw A-scans to alleviate these limitations, but their effectiveness is hindered by the conservative sampling rates and the uniform modeling of flow and background signals. In this study, we introduce a novel blood flow-aware network, named ASBA (A-line ROI State space model and B-line phase Attention), to reconstruct ODT images from highly sparsely sampled raw A-scans. Specifically, we propose an A-line ROI state space model to extract sparsely distributed flow features along the A-line, and a B-line phase attention to capture long-range flow signals along each B-line based on phase difference. Moreover, we introduce a flow-aware weighted loss function that encourages the network to prioritize the accurate reconstruction of flow signals. Extensive experiments on real animal data demonstrate that the proposed approach clearly outperforms existing state-of-the-art reconstruction methods.

Figures (11)

• Figure 1: Examples of sparse ODT reconstruction. Due to big gaps among sampled A-lines at high sparsity, traditional pipeline zhao2000phase that estimates flow speed directly from phase differences of adjacent A-lines fails to reconstruct flow images. The state-of-the-art (SOTA) sparse ODT reconstruction method li2025sparse also loses most weak flow signals in small vessels, while our proposed ASBA effectively preserves most flows in the reconstructed image.
• Figure 2: Our sparse ODT reconstruction system. The raw B-scan is highly downsampled with a large stride along the B-line. The sparse magnitude and phase responses are forwarded to the proposed ASBA network to reconstruct the B-scan image.
• Figure 3: Network Blocks. (a) Magnitude Module, (b) Fusion Block, (c) Phase Module, (d) A-line ROI State Space Model (A-RS) Block, and (e) B-line Phase Attention (B-PA) Block.
• Figure 4: Operations in the B-PA Block to map the phase difference feature back to the phase domain.
• Figure 5: Qualitative results of $\times 16$ sparsity on MCD-AW dataset. Regions of interest are zoomed in.