Automatic segmentation of fluorescence lifetime microscopy images of cells using multiresolution community...

Area of Science:

• Biomedical Imaging
• Computational Biology
• Microscopy Techniques

Background:

• Fluorescence Lifetime Imaging Microscopy (FLIM) provides valuable functional information about biological samples.
• Accurate segmentation of FLIM images is crucial for quantitative analysis of cellular structures and processes.
• Existing segmentation methods may struggle with noise and achieving consistent accuracy across different resolutions.

Purpose of the Study:

• To propose and evaluate an automatic method for segmenting fluorescence lifetime (FLT) imaging microscopy (FLIM) images.
• To adapt a multiresolution community detection network segmentation approach for FLIM image analysis.
• To compare the performance of the proposed method against a spectral clustering-based method.

Main Methods:

• Framed FLIM image segmentation as identifying segments with distinct average FLTs against background.
• Employed a multiresolution network segmentation where pixels are nodes and FLT similarity defines edges.
• Investigated the impact of network resolution on segmentation accuracy and segment size.

Main Results:

• The proposed method successfully segmented FLIM images, with lower resolution yielding larger segments and higher resolution yielding smaller segments.
• Mean-square error in FLT segment estimation consistently decreased with increasing network resolution.
• The multiresolution community detection method outperformed spectral clustering, which produced noisy segments and lacked consistent error reduction.

Conclusions:

• The developed automatic multiresolution community detection method offers an effective approach for FLIM image segmentation.
• This method demonstrates superior performance and robustness compared to spectral clustering for FLIM data.
• The findings highlight the potential of network-based segmentation for advancing quantitative FLIM image analysis.