Extraction of target specimens from bioholographic images using interactive graph cuts

Area of Science:

• Biomedical Imaging
• Computational Biology
• Microscopy

Background:

• Digital holographic microscopy is increasingly used for analyzing transparent biological specimens.
• Accurate segmentation of target specimens is crucial for subsequent analysis like identification and tracking.
• Existing methods may struggle with specimens exhibiting weak or indistinct boundaries.

Purpose of the Study:

• To develop and present an interactive graph cuts approach for segmenting target specimens in bioholographic images.
• To offer a robust method capable of handling challenging segmentation scenarios, including weak boundaries and multiple similar objects.

Main Methods:

• An interactive graph cuts algorithm was employed for image segmentation.
• The method integrates both regional and boundary information to enhance segmentation accuracy.
• A user interface was developed for intuitive foreground/background differentiation, with a dynamically adjustable coefficient for weighting regional and boundary information.

Main Results:

• The interactive graph cuts method demonstrated robust performance in segmenting various target specimens from bioholographic images.
• The technique proved effective even when dealing with multiple similar objects or very weak object boundaries.
• Comparative analysis against a level-set-based method confirmed the efficacy of the proposed approach.

Conclusions:

• The interactive graph cuts technique provides an effective and robust solution for specimen segmentation in bioholographic imaging.
• This method enhances the capability for high-level analysis of biological specimens using digital holographic microscopy.
• The approach offers a valuable tool for researchers working with challenging bioholographic image data.