Multiphoton fluorescent images with a spatially varying background signal: a ML deconvolution method

Area of Science:

• Neuroscience
• Biomedical Imaging
• Optical Microscopy

Background:

• Multiphoton laser scanning microscopy enables deep brain imaging.
• Optical images, including multiphoton fluorescent images, suffer from degradation due to light scattering, absorption, laser instability, and staining issues.
• Image degradation can be modeled as stochastic noise and background signal.

Purpose of the Study:

• To extend the split-gradient deconvolution method (SGM) by incorporating a maximum-likelihood estimation for spatially varying background signals.
• To address the limitation of the previous SGM which assumed a constant background.
• To evaluate the performance of the extended SGM on synthetic and real multiphoton fluorescent images.

Main Methods:

• The study extends the split-gradient deconvolution method (SGM).
• A maximum-likelihood estimation step is added to determine a spatially varying background signal.
• The performance is evaluated using synthetic and actual multiphoton fluorescent images.

Main Results:

• The assumption of a constant background signal is not always valid in multiphoton laser microscopy.
• The proposed method, incorporating spatially varying background estimation, demonstrates improved accuracy.
• The extended SGM shows enhanced performance compared to the original SGM algorithm.

Conclusions:

• The developed method effectively handles spatially varying background signals in multiphoton microscopy.
• This advancement improves the quality and accuracy of deconvolution for deeper brain imaging.
• The findings contribute to more reliable neuroscientific investigations using advanced microscopy techniques.