Automated Stabilization, Enhancement and Capillaries Segmentation in Videocapillaroscopy

Area of Science:

• Biomedical Engineering
• Medical Imaging
• Microcirculation Analysis

Background:

• Oral capillaroscopy is vital for non-invasive microcirculation assessment in diagnosing pathologies.
• Current videocapillaroscopy in the oral cavity faces limitations in resolution, magnification, and operator influence.
• Anatomical-pathological lesions are key factors in disease progression observed via capillaroscopy.

Purpose of the Study:

• To develop a computerized support system to overcome limitations in oral capillaroscopy.
• To enhance the reliability and accuracy of microcirculation analysis in oral mucosal video sequences.
• To improve the diagnostic and prognostic capabilities of oral capillaroscopy.

Main Methods:

• A computerized system was developed for stabilization, enhancement, and automatic capillary segmentation.
• Stabilization utilized a seed point coupling method within a classification process.
• Enhancement involved temporal analysis of frames, and segmentation employed U-net deep network transfer learning.

Main Results:

• The system achieved a Jaccard index of 90.1% and an accuracy of 96.2% in capillary segmentation.
• Quantitative assessment confirmed significant signal improvement through the developed techniques.
• The method demonstrated high effectiveness in analyzing oral capillaroscopy data.

Conclusions:

• The developed computerized system significantly improves oral capillaroscopy analysis.
• This technology can assist in diagnosing and monitoring conditions affecting microcirculation, like rheumatologic and cardiovascular disorders.
• The findings support the broader clinical adoption of advanced oral capillaroscopy techniques.