Synthesis of cervical tissue second harmonic generation images using Markov random field modeling

Area of Science:

• Biomedical Imaging
• Computational Biology
• Materials Science

Background:

• Second harmonic generation (SHG) imaging provides structural information of biological tissues.
• Cervical tissue structure, particularly fiber and pore distribution, is crucial for understanding pregnancy.
• Accurate modeling of cervical tissue morphology is needed for advanced analysis.

Purpose of the Study:

• To develop a statistical image modeling approach for synthesizing cervical tissue SHG images.
• To generate realistic binary images representing fiber and pore areas.
• To validate the synthesized images using classification of SHG data from different pregnancy stages.

Main Methods:

• Image processing pipeline: noise removal, contrast enhancement, and optimal thresholding to obtain binary images.
• Markov random field (MRF) modeling of fiber and pore areas.
• Parameter estimation using the least squares method for MRF model.

Main Results:

• Successful synthesis of binary images representing cervical tissue fiber and pore areas.
• Demonstrated effectiveness of the synthesis through classification of SHG images from normal pregnancy stages.
• Generated images exhibit realistic structural features.

Conclusions:

• The developed MRF-based statistical image modeling approach effectively synthesizes cervical tissue binary images.
• The synthesized images are suitable for studying cervical tissue morphology during pregnancy.
• This method facilitates the generation of realistic data for computational analysis in reproductive biology.