Image-guided filtering for improving photoacoustic tomographic image reconstruction

Area of Science:

• Medical Imaging
• Biomedical Optics
• Image Reconstruction

Background:

• Photoacoustic imaging relies on reconstruction algorithms that often prioritize single image features like smoothness or sharpness.
• Existing methods may not optimally balance multiple desired features in reconstructed images.
• Postprocessing techniques are needed to refine results from standard reconstruction algorithms.

Purpose of the Study:

• To develop and evaluate a guided filtering approach for enhancing photoacoustic image reconstruction.
• To combine desirable image features (e.g., sharpness and smoothness) in the reconstructed output.
• To improve the signal-to-noise ratio (SNR) of photoacoustic images using a computationally efficient method.

Main Methods:

• A guided filtering approach was implemented as a postprocessing step for Tikhonov and total variational regularization methods.
• Linear backprojection (LBP) results were utilized as the guiding image for the filtering process.
• The method was validated using both numerical and experimental phantom data.

Main Results:

• The guided filtering approach significantly improved the signal-to-noise ratio (SNR) of reconstructed photoacoustic images, with improvements up to 11.23 dB.
• The proposed method demonstrated computational efficiency compared to other techniques.
• Performance evaluation showed the guided filtering approach outperformed state-of-the-art basis pursuit deconvolution and standard denoising methods.

Conclusions:

• Guided filtering offers an effective postprocessing strategy to enhance photoacoustic image reconstruction quality.
• The method successfully integrates multiple image features, leading to improved SNR and visual clarity.
• This computationally efficient technique presents a valuable advancement for photoacoustic image analysis.