Fluorescence in Situ Hybridization Cell Image Segmentation Method

Area of Science:

• Molecular Cytogenetics
• Biomedical Imaging
• Computational Biology

Background:

• Fluorescence in situ hybridization (FISH) is crucial for diagnosing cancers and genetic disorders at the cellular level.
• Accurate cell segmentation in FISH images is essential for quantifying DNA/RNA signals but is challenging due to image quality issues.
• Low contrast, unclear boundaries, and cell defects in FISH images impede reliable cell identification and segmentation.

Purpose of the Study:

• To develop and validate an advanced cell image segmentation method specifically for fluorescence in situ hybridization (FISH) applications.
• To address the limitations of existing methods in handling poor image quality in FISH datasets.
• To improve the accuracy and reliability of cell segmentation for downstream diagnostic analyses.

Main Methods:

• The proposed method utilizes an improved adaptive histogram equalization technique for image enhancement.
• A watershed segmentation approach is employed, using optimal segmentation feature values as seed points.
• The method was applied to segment leukemia cells from fluorescence in situ hybridization images.

Main Results:

• The developed method demonstrated high performance in segmenting 273 leukemia cells across 5 images.
• Achieved a remarkable detection rate of 99% and an effectiveness of 94.5%.
• Successfully addressed challenges posed by low contrast and unclear cell boundaries in FISH images.

Conclusions:

• The proposed fluorescence in situ hybridization cell image segmentation method accurately segments cells, proving effective for diagnostic applications.
• The technique offers a high detection rate and effectiveness, improving the reliability of FISH-based cellular analysis.
• This advancement facilitates more precise quantification of genetic material within cells for cancer and genetic disease diagnosis.