Resolution Enhancement of Under-sampled Photoacoustic Microscopy Images using Implicit Neural Representations

TL;DR

This work proposes an approach based on Implicit Neural Representations (INR) that learns a continuous mapping from spatial coordinates to initial acoustic pressure, overcoming the limitations of discrete imaging and enhancing AR-PAM's resolution.

Abstract

Acoustic-Resolution Photoacoustic Microscopy (AR-PAM) is promising for subcutaneous vascular imaging, but its spatial resolution is constrained by the Point Spread Function (PSF). Traditional deconvolution methods like Richardson-Lucy and model-based deconvolution use the PSF to improve resolution. However, accurately measuring the PSF is difficult, leading to reliance on less accurate blind deconvolution techniques. Additionally, AR-PAM suffers from long scanning times, which can be reduced via down-sampling, but this necessitates effective image recovery from under-sampled data, a task where traditional interpolation methods fall short, particularly at high under-sampling rates. To address these challenges, we propose an approach based on Implicit Neural Representations (INR). This method learns a continuous mapping from spatial coordinates to initial acoustic pressure, overcoming the limitations of discrete imaging and enhancing AR-PAM's resolution. By treating the PSF as a learnable parameter within the INR framework, our technique mitigates inaccuracies associated with PSF estimation. We evaluated our method on simulated vascular data, showing significant improvements in Peak Signal-to-Noise Ratio (PSNR) and Structural Similarity Index (SSIM) over conventional methods. Qualitative enhancements were also observed in leaf vein and in vivo mouse brain microvasculature images. When applied to a custom AR-PAM system, experiments with pencil lead demonstrated that our method delivers sharper, higher-resolution results, indicating its potential to advance photoacoustic microscopy.

Figures (8)

• Figure 1: The process of generating low-sampling images from the full-scanning ones.
• Figure 2: The pipeline of proposed the sparse deconvolution reconstruction method. The network takes as input the coordinates $p$ of densely sampled high-resolution images and predicts the image intensities $I$ at these locations. Subsequently, PSF convolution and down-sampling operations are performed. Finally, the network is optimized by minimizing the loss between the predicted sparsely sampled AR-PAM images and the actual acquired AR-PAM images.
• Figure 3: Qualitatively compare the deblurring and sparse reconstruction performance of our method with that of blind deconvolution using different interpolation techniques. Examples of simulated vascular images at 2x down-sampling are provided.
• Figure 4: Qualitatively compare the deblurring and sparse reconstruction performance of our method with that of blind deconvolution using different interpolation techniques. Examples of simulated vascular images at 4x down-sampling are provided.
• Figure 5: Qualitatively compare the deblurring and sparse reconstruction performance of our method with that of blind deconvolution using different interpolation techniques. Examples of leaf vein data at 2x down-sampling. (a) by bilinear+blind deconvolution, (b) by bicubic+blind deconvolution, (c) by our method, (d) OR-PAM(Ground truth).