A dicentric chromosome identification method based on clustering and watershed algorithm.

Area of Science:

• Cytogenetics
• Radiation Biology
• Image Analysis

Background:

• Microscopic dicentric chromosome identification is crucial for biological dosimetry but faces efficiency challenges.
• Automated methods are needed to improve the accuracy and speed of this process.

Purpose of the Study:

• To develop and evaluate a novel joint processing algorithm for efficient and accurate dicentric chromosome identification.
• To assess the algorithm's performance in dose estimation compared to traditional methods.

Main Methods:

• A joint clustering and watershed algorithm was employed for segmenting and identifying individual chromosomes from microscopic images.
• The algorithm's true positive rate and positive predictive value were evaluated using specific parameters (m=1, n=1).
• Dose estimation accuracy was assessed based on the number of identified dicentric chromosomes per cell.

Main Results:

• The proposed algorithm achieved a true positive rate of 76.6% and a positive predictive value of 76.6% for dicentric chromosome identification in high-dose samples.
• These results surpass the performance of a threshold algorithm (63.9% true positive rate, 63.5% positive predictive value).
• A dose estimation pass rate of 80% was achieved using 500 cells in high-dose scenarios, with recommendations for increased cell counts in low-dose estimations.

Conclusions:

• The joint clustering and watershed algorithm offers a significant improvement in dicentric chromosome identification efficiency and accuracy.
• This method provides a reliable basis for radiation dose estimation, particularly in high-dose exposures.
• Further optimization, including analyzing more cells, is recommended to enhance dose estimation accuracy for low-dose exposures.