Matrix-Free Photoacoustic Image Reconstruction via Sensor-Token Self-Attention

2026-07-28

AI Summary

This work addresses the ill-posed inverse problem of initial pressure reconstruction in photoacoustic tomography under sparse-view conditions, where conventional methods rely on system matrices that incur high computational costs and hinder real-time application. To overcome this limitation, the authors propose the Sensor Attention Network (SAN), which introduces a matrix-free Transformer architecture to the field for the first time. SAN treats the full temporal signal from each sensor as a token and leverages self-attention mechanisms to enable end-to-end image reconstruction directly from raw measurements. By integrating k-space H-matrix generation, vessel-weighted loss, and data augmentation, the method achieves high-fidelity reconstructions (SSIM = 0.522, PSNR = 22.09 dB, NMSE = 0.233) while accelerating inference by at least one order of magnitude compared to ISTA, SBTV, and LISTA, thereby enabling real-time clinical deployment.

Abstract

Photoacoustic tomography (PAT) combines the optical absorption contrast of biological tissue with the spatial resolution of ultrasound, yet recovering the initial pressure distribution from sparse-view sensor measurements remains an ill-posed inverse problem. Iterative compressive-sensing solvers and unrolled deep networks both retain a dependence on the system matrix at inference, which leaves real-time clinical reconstruction computationally expensive. This paper proposes the Sensor Attention Network (SAN), a Transformer-based architecture that treats the full time series of each sensor as a token and maps raw measurements directly to the reconstructed image without invoking the system matrix at inference. For training and benchmarking, an analytical k-space H-matrix is constructed and validated against the k-Wave pseudo-spectral solver under matched geometry, achieving a mean per-sensor Pearson correlation of 0.919 +/- 0.049, with k-space apodization and Gaussian temporal damping acting synergistically to reduce the energy-normalized mismatch by 49%. Trained with a vessel-weighted loss on 488 augmented samples and evaluated on 46 held-out samples against ISTA, split-Bregman total variation (SBTV), and learned ISTA (LISTA), SAN attains the highest mean SSIM (0.522) and PSNR (22.09 dB) and the lowest NMSE (0.233). Paired t-tests and Wilcoxon signed-rank tests confirm the superiority of SAN over LISTA on PSNR, NMSE, and Pearson correlation at p < 1e-8, and over ISTA and SBTV on all fidelity metrics. By bypassing the H-matrix at inference, SAN reduces reconstruction time by at least an order of magnitude, supporting real-time PAT reconstruction.

Problem

• Photoacoustic tomography
• ill-posed inverse problem
• sparse-view reconstruction
• real-time imaging
• system matrix dependency

Innovation

• matrix-free reconstruction
• sensor-token self-attention
• Transformer-based photoacoustic imaging
• real-time PAT
• k-space H-matrix