Model-based quantitative photoacoustic tomography with directional total variation.
Jiaming LiuLi QiYanqiu FengQiugen HuShuangyang ZhangPublished in: Journal of biophotonics (2024)
In photoacoustic tomography (PAT), acoustic inversion aims to recover the spatial distribution of light energy deposition within the imaging object from the signals captured by detectors. To achieve quantitative imaging, optical inversion is further employed to derive absorption coefficient (AC) images. However, limitations such as restricted detection angles and inherent noise lead to substantial artifacts and degradation in the quality of PAT images, consequently affecting the accuracy of optical inversion results. In this study, we propose a directional total variation constrained optical inversion model to reconstruct the AC image. By incorporating anatomy prior information into the optical inversion process, our method can effectively suppress artifacts in AC images while maintaining structural integrity. Simulation, phantom, and in vivo experimental results demonstrate that our method significantly improves the reconstructed AC image quality. Our method provides a reliable foundation for achieving high-quality quantitative PAT imaging.
Keyphrases
- high resolution
- image quality
- deep learning
- contrast enhanced
- high speed
- convolutional neural network
- mass spectrometry
- computed tomography
- optical coherence tomography
- magnetic resonance imaging
- fluorescence imaging
- healthcare
- dual energy
- working memory
- magnetic resonance
- air pollution
- machine learning
- health information
- quality improvement
- quantum dots
- label free