微血管光声三维重建与剪切波粘弹性估计3D photoacoustic reconstruction and shear-wave viscoelastic estimation of microvasculature
微血管的几何构型与力学特性，对肿瘤、糖尿病等疾病的早期诊断有参考价值。但目前缺少能在真值已知的条件下，系统验证成像与粘弹性反演算法的仿真平台。我搭建了光声（PA）与剪切波（SW）双模态联合仿真框架，在受控条件下评估这些算法的表现。
The geometry and mechanics of microvessels carry information for early diagnosis of diseases such as cancer and diabetes, yet there is no simulation platform that can systematically test imaging and viscoelastic inversion algorithms against a known ground truth. I built a joint photoacoustic (PA) and shear-wave (SW) simulation framework to evaluate these algorithms under controlled conditions.
方法Method
- 三维微血管建模：用骨架-体素场方法生成单管、随机网络和真实网络三类模型。3D vessel modelling: skeleton-voxel field models of a single vessel, a random network and a real network.
- 多物理场仿真：在 COMSOL 中分别设置压力声学瞬态模块（PA）和固体力学模块（SW）。Multiphysics simulation: transient pressure acoustics (PA) and solid mechanics (SW) in COMSOL.
- 传感器阵列：PA 为四侧 256 路线阵，SW 为底面 64 路近场线阵。Sensor arrays: a four-sided 256-channel linear array for PA, a 64-channel near-field array at the bottom for SW.
- 光声重建：用 k-Wave 时间反演算法重建三维初始声压分布。Photoacoustic reconstruction: 3D initial pressure recovered with k-Wave time reversal.
- 粘弹性估计：滑窗互相关渡越时间法，配合 Maxwell 模型估计剪切波速、模量和粘度。Viscoelastic estimation: sliding-window cross-correlation transit time with a Maxwell model to estimate shear-wave speed, modulus and viscosity.
结果Results
- PA 重建（随机网络）PA reconstruction
- PSNR 9.70 dB，SSIM 0.72 （ROI 区域）(ROI)
- 剪切波速Shear-wave speed
- 3.077 m/s 相对误差 0.76%relative error 0.76%
- 剪切模量 GShear modulus G
- 9.48 kPa 相对误差 5.87%relative error 5.87%
- 粘度 ηViscosity η
- 0.0534 Pa·s 相对误差 6.3%relative error 6.3%
- 噪声鲁棒性Noise robustness
- 在无噪声、40、30、20 dB 四级信噪比下，PA 重建的 SSIM 波动小于 0.01；波速误差在 30 dB 及以上低于 1.5%，20 dB 时约 2.16%。Across noise-free, 40, 30 and 20 dB conditions, PA reconstruction SSIM varied by less than 0.01; wave-speed error stayed below 1.5% at 30 dB and above, and was about 2.16% at 20 dB.
以上均为仿真结果，尚未经过实验验证。All results are from simulation and have not yet been validated experimentally.