HeLa cell segmentation using digital image processing.

Area of Science:

• Biomedical Imaging
• Computational Biology
• Cell Biology

Background:

• Accurate cell segmentation is critical for cancer research, often relying on resource-intensive deep learning methods.
• Traditional image processing offers a viable, low-resource alternative for cell segmentation, especially for cell lines like HeLa.
• Serial Block-Face Scanning Electron Microscopy (SBF-SEM) provides high-resolution imaging data crucial for detailed cellular analysis.

Purpose of the Study:

• To develop and validate a digital image processing algorithm for segmenting the nucleus and overall shape of HeLa cells.
• To provide an efficient and accessible method for cell segmentation, overcoming the limitations of deep learning approaches.
• To enable accurate identification of cellular morphology for disease analysis and diagnostic purposes.

Main Methods:

• Utilized 300 SBF-SEM images of HeLa cells.
• Employed morphological erosion for cell separation and distance calculation to identify the central cell.
• Restored cell shape using the eroded form and segmented the nucleus using parameter-based verification stages.

Main Results:

• The proposed algorithm successfully segmented HeLa cell shapes and nuclei.
• Comparative analysis using four similarity metrics demonstrated superior performance over an alternative algorithm.
• The algorithm proved effective in selecting the central cell and accurately identifying nuclear features.

Conclusions:

• The developed image processing algorithm is a significant advancement for accurate disease analysis through cell segmentation.
• This method facilitates the measurement of cell shapes and the detection of morphological alterations or organelle damage.
• The algorithm presents a practical tool for research and potentially for diagnostic applications in cell-based studies.