Threshold estimation based on local minima for nucleus and cytoplasm segmentation

Area of Science:

• Medical Imaging
• Computational Biology
• Image Processing

Background:

• Image segmentation is crucial for analyzing image data.
• Accurate segmentation of white blood cells (WBCs) is essential for hematological analysis.
• Existing methods may face challenges with complex cellular structures and overlapping cells.

Purpose of the Study:

• To develop an automated method for segmenting nuclei and cytoplasm in white blood cells (WBCs).
• To improve the accuracy and efficiency of cellular component segmentation in medical images.
• To provide a robust solution for segmenting WBCs even when adjacent to red blood cells (RBCs).

Main Methods:

• A novel segmentation approach based on histogram analysis and local minima detection.
• Initial thresholding using image intensity statistics.
• Post-processing techniques including Otsu's method and morphological operations for nucleus and cytoplasm segmentation.
• Integration of SLIC and watershed methods for complex segmentation scenarios.

Main Results:

• The proposed method successfully segmented nuclei and cytoplasm in WBCs.
• Performance was validated on two public datasets, showing competitive results against state-of-the-art methods.
• The technique demonstrated effectiveness in handling images with varying cell separation and proximity to RBCs.

Conclusions:

• The developed method provides a reliable approach for segmenting WBC nuclei and cytoplasm.
• The technique's foundation in histogram approximation and local minima offers a robust segmentation strategy.
• The method shows significant utility and satisfactory performance in automated cell image analysis.