A Novel Technique for Fluorescence Lifetime Tomography

Area of Science:

• Biomedical Optics
• Medical Imaging
• Deep Learning Applications

Background:

• Fluorescence lifetime imaging (FLI) offers quantitative insights into the molecular environment of diseased tissues.
• Current 2D FLI is mature, but 3D imaging in deep, scattering tissues remains a significant challenge.
• Scattering in biological tissues severely limits the depth and resolution of traditional FLI.

Purpose of the Study:

• To develop a deep learning (DL) approach for accurate 3D fluorescence lifetime and intensity reconstructions in deep tissues.
• To overcome the limitations imposed by scattering in deep tissue imaging.
• To enable mesoscopic and macroscopic scale molecular imaging using FLI.

Main Methods:

• A novel deep neural network, AUTO-FLI, was developed for 3D reconstruction.
• An *in silico* scheme was implemented for accurate generation of fluorescence lifetime data.
• The DL model was rigorously validated using both simulated data and experimental phantoms.

Main Results:

• AUTO-FLI successfully performed 3D intensity and quantitative lifetime reconstructions in highly scattering media.
• The method demonstrated accurate quantitative estimations of both intensity and lifetime distributions.
• Validation confirmed the model's reliability in challenging deep tissue environments.

Conclusions:

• AUTO-FLI significantly advances 3D fluorescence lifetime imaging capabilities in deep tissues.
• The developed DL approach overcomes scattering limitations, paving the way for mesoscopic and macroscopic FLI.
• This technology holds substantial potential for molecular imaging of diseased tissues at greater depths and scales.