Cell nuclei and cytoplasm joint segmentation using the sliding band filter

Area of Science:

• Cell biology
• Bioimaging
• Computer vision

Background:

• Microscopy cell image analysis is crucial for biological research, particularly multivariate fluorescence microscopy.
• Manual cell analysis is subjective and inefficient, necessitating automated methods for large-scale studies.
• Traditional automated cell analysis relies on image segmentation, which is challenging and sensitive to image quality.

Purpose of the Study:

• To develop a novel, robust, and automated approach for cell detection and shape estimation in multivariate microscopy images.
• To overcome limitations of traditional image segmentation methods in cell analysis.
• To improve the accuracy and objectivity of cell image analysis in biological research.

Main Methods:

• A new cell detection and shape estimation method based on the sliding band filter (SBF) was developed.
• The SBF approach leverages intuitive parameters related to cell size for detecting cell nucleus and cytoplasm.
• Cytoplasm shape estimation is guided by nuclear detections, with incorporated overlap correction and shape regularization.

Main Results:

• The SBF method demonstrated effective cell detection and shape estimation on simulated and real biological datasets.
• On the Drosophila melanogaster dataset, nuclei detection achieved 95%/69% precision/recall, and cytoplasm detection achieved 82%/90% precision/recall.
• The overall accuracy for cell detection and shape estimation reached 76% on the challenging Drosophila dataset.

Conclusions:

• The sliding band filter (SBF) offers an effective and intuitive method for automated cell detection and shape estimation in complex microscopy images.
• This approach enhances objectivity and efficiency in cell image analysis, supporting large-scale biological studies.
• The method shows promise for advancing quantitative cell biology research by providing reliable cell morphology data.