Two-Dimensional Plane for Multi-Scale Quantification of Corneal Subbasal Nerve Tortuosity

Area of Science:

• Ophthalmology
• Medical Imaging
• Computational Biology

Background:

• Vessel tortuosity is a key indicator in various medical conditions.
• Accurate and objective assessment of tortuosity is crucial for diagnosis and monitoring.
• Current methods for tortuosity analysis can be subjective and time-consuming.

Purpose of the Study:

• To develop and evaluate a novel automated system for estimating and interpreting vascular tortuosity.
• To identify optimal tortuosity measures and representations for accurate agreement with human observers.
• To introduce a new visualization tool, the tortuosity plane (TP), for enhanced interpretation.

Main Methods:

• A supervised learning strategy was used, driven by expert grading of 140 corneal nerve images.
• Eighteen tortuosity measures, including curvature and inflection point density at multiple scales, were investigated.
• A tortuosity plane (TP) was developed, mapping images based on selected tortuosity measures for interpretation.

Main Results:

• The optimal tortuosity representation combined mean curvature at spatial scales 2 and 5.
• The automated system demonstrated strong agreement with expert observers, both globally and per-level (Spearman's correlation, P < 0.0001).
• The system's accuracy surpassed at least one expert in 2 out of 3 comparisons, with significant improvements (P < 0.001).

Conclusions:

• The automated system effectively stratifies images into four tortuosity levels, matching or exceeding expert accuracy.
• The tortuosity plane (TP) enables continuous, two-dimensional assessment, allowing for finer discrimination of image tortuosity.
• This automated approach offers a more objective and potentially more accurate method for analyzing vascular tortuosity.