Random Forests for Dura Mater Microvasculature Segmentation Using Epifluorescence Images

Area of Science:

• Medical Imaging
• Computational Biology
• Biomedical Engineering

Background:

• Accurate segmentation of microvascular structures is crucial for understanding tissue physiology and disease.
• Quantitative analysis of vessel remodeling requires precise characterization of dural microvasculature.

Purpose of the Study:

• To develop an automated method for segmenting thin vessel structures in biological tissues.
• To improve the accuracy of microvascular segmentation compared to existing methods.

Main Methods:

• A supervised random forest (RF) classifier was employed for vessel segmentation.
• Multiscale features, including Hessian, oriented second derivatives, Laplacian of Gaussian, and line features, were utilized.
• A multiscale line detector feature was incorporated to enhance detection of faint vessels.

Main Results:

• The RF classifier demonstrated superior performance in segmenting thin vessel structures.
• The proposed method achieved approximately 20% and 25% improvement over Niblack and Otsu threshold-based methods, respectively.
• Experimental results on epifluorescence imagery validated the effectiveness of the RF approach.

Conclusions:

• The developed RF-based method offers a robust and accurate solution for automatic microvascular segmentation.
• This technique facilitates quantitative analysis of physiological changes in tissues.
• The improved detection of faint vessels aids in comprehensive microvascular characterization.