Deconvolution for multimode fiber imaging: modeling of spatially variant PSF.

Area of Science:

• Biomedical Optics
• Microscopy Imaging
• Optical Engineering

Background:

• Minimally-invasive endoscopy using multimode optical fibers (MMF) faces challenges due to spatially variant point spread functions (PSFs).
• Standard deconvolution algorithms struggle with non-uniform PSFs, limiting image quality and effective field of view (FOV).

Purpose of the Study:

• To develop a method for compensating spatially variant PSFs in MMF imaging.
• To enable effective deconvolution and image enhancement for MMF-based endoscopy.

Main Methods:

• Developed svmPSF, an open-source Java framework for ImageJ.
• Utilized principal component analysis (PCA) on point response measurements to model the spatially variant PSF.
• Applied a modified Richardson-Lucy deconvolution algorithm with the generated PSF model.

Main Results:

• Successfully modeled spatially variant PSFs from MMF measurements.
• Deblurred and regularized fluorescence images of biological samples (beads, neurons).
• Demonstrated effective increase in the usable FOV for MMF imaging.

Conclusions:

• The svmPSF framework effectively models and compensates for spatially variant PSFs in MMF imaging.
• This approach significantly improves image quality and extends the FOV in minimally-invasive endoscopic applications.
• svmPSF offers a practical solution for enhancing MMF-based microscopy and endoscopy.