Volumetric segmentation of biological cells and subcellular structures for optical diffraction tomography images

Area of Science:

• Biomedical research
• Quantitative imaging
• Cell biology

Background:

• Optical Diffraction Tomography (ODT) provides 3D quantitative imaging of cells and internal structures.
• Accurate 3D segmentation of ODT images is crucial for quantitative analysis but challenging with current methods.

Purpose of the Study:

• To present ODT-SAS, a novel semi-automatic method for segmenting cells, nucleoli, and lipid structures (LS) in ODT images.
• To evaluate the performance of ODT-SAS against existing segmentation techniques.

Main Methods:

• ODT-SAS utilizes a combination of non-machine-learning techniques for segmentation.
• Comparative analysis was performed against Cellpose for cell segmentation and manual slice-by-slice segmentation for organelle segmentation.

Main Results:

• ODT-SAS demonstrated superior performance compared to Cellpose in cell segmentation.
• The method shows significant potential for accurate detection of cells, nucleoli, and lipid structures.

Conclusions:

• ODT-SAS offers an effective solution for accurate 3D segmentation in ODT imaging.
• This technique enhances quantitative analysis capabilities in biomedical research involving cell and organelle imaging.