Deep system prior based graph convolution network for NIR-II fluorescence molecular tomography

Area of Science:

• Biomedical Imaging
• Medical Physics
• Computational Imaging

Background:

• Fluorescence molecular tomography (FMT) is valuable for early-stage tumor detection.
• Severe photon scattering in FMT leads to ill-posed inverse problems, hindering accuracy and morphological reconstruction.
• Current FMT methods struggle to meet practical requirements for precise tumor imaging.

Purpose of the Study:

• To enhance the efficiency and accuracy of FMT.
• To improve the morphological performance of FMT reconstruction.
• To address the ill-posed nature of FMT caused by photon scattering.

Main Methods:

• Utilized second near-infrared (NIR-II) fluorescence imaging to reduce tissue scattering.
• Developed a deep system prior based graph convolution network (DSPGN) for FMT.
• Incorporated system spatial priors and graph structures into the reconstruction process.

Main Results:

• DSPGN demonstrated superior performance in location accuracy compared to existing methods.
• The proposed method significantly improved shape recovery capability in FMT.
• Both numerical simulations and in vivo experiments validated DSPGN's effectiveness.

Conclusions:

• DSPGN effectively recovers both location and morphology of fluorescence sources.
• The method shows potential for advancing FMT applications in NIR-II imaging.
• DSPGN offers a promising solution for overcoming FMT's inherent challenges.