Three-dimensional Vision-based Nail-fold Morphological and Hemodynamic Analysis

Abstract

In this paper, a Three-dimensional Vision-based Nail-fold Morphological and Hemodynamic Analysis (TVNMHA) is proposed to automatically extract morphological/hemodynamic features from an image sequence, reconstruct the corresponding three-dimensional vacular tubular models, and visualize the dynamic blood flow in the model constructed. The morphological features extracted include number, density, width/height, width of the curved segment, arteriolar limb caliber, curved segment caliber, venular limb caliber, blood color and tortuosity of capillaries. The corresponding pathological information derived has a spatial precision accurate up to 1.6£gm. TVNMHA also employs optical flow method (OFM) to measure the velocity of the flowing plasma gaps between two consecutive frames with an error 10~32£gm/sec [1]. The laser Doppler velocimetry is a common tool employed in measuring capillary with larger diameter (is about 500£gm). The average velocity of an area measured comes with an error around 100£gm/sec [2], much higher than that of TVNMHA. TVNMHA can separately derive velocities for each capillaries within the whole microscopic range and obtain diversified morphological/hemodynamic features of capillaries with a low-cost equipment setup.