Related Experiment Video
Updated: Aug 14, 2025
14:09
High-Throughput Total Internal Reflection Fluorescence and Direct Stochastic Optical Reconstruction Microscopy Using a Photonic Chip
Published on: November 16, 2019
7.0K
High-fidelity mesoscopic fluorescence molecular tomography based on SSB-Net
Optics Letters
|January 13, 2023
Summary
We developed a spatially adaptive split Bregman network (SSB-Net) to improve mesoscopic fluorescence molecular tomography (MFMT) imaging. This novel deep learning approach enhances reconstruction fidelity for multifluorophore imaging in reflective geometry.
Area of Science:
- Biomedical Optics
- Medical Imaging
- Fluorescence Molecular Tomography
Background:
- Mesoscopic fluorescence molecular tomography (MFMT) in reflective geometry faces challenges with spatial sensitivity nonuniformity and ill-posed reconstruction.
- These limitations hinder accurate imaging of fluorescent targets within scattering media.
Purpose of the Study:
- To introduce a novel deep learning framework, the spatially adaptive split Bregman network (SSB-Net), to address MFMT imaging fidelity issues.
- To simultaneously compensate for spatial sensitivity variations and promote sparse reconstructions in MFMT.
Main Methods:
- The SSB-Net is derived by unfolding the split Bregman algorithm, integrating residual blocks and 3D convolutional neural networks (3D-CNNs).
- Each layer adaptively learns spatially nonuniform error compensation, spatially dependent proximal operators, and sparsity transformations.
Main Results:
- Simulations and experimental results demonstrate high-fidelity MFMT reconstruction of multifluorophores at various positions.
- The method successfully imaged targets within a depth of a few millimeters.
Conclusions:
- The proposed SSB-Net effectively overcomes key limitations in reflective MFMT, enabling accurate reconstruction.
- This work advances reflection-mode diffuse optical imaging towards practical clinical applications.
