Quantitative Nail Fold Capillary Blood Flow Using Capillaroscopy System and ImageJ Software in Healthy Individuals

Abstract

Purpose: Measuring the blood flow velocity in capillaries is a useful method for diagnosing many diseases. Despite increasing interest in nailfold capillaroscopy, objective measures of capillary structure and blood flow have been rarely studied. This study aimed to measure the blood flow velocity along the capillary central line using capillaroscopy system, and also ImageJ software used Scale-Invariant Feature Transform (SIFT) tracking algorithms and Kalman filter for image processing. Materials and Methods: The Red Blood Cells (RBCs) velocity in capillaries of finger nailfold was measured in 12 normal cases using a novel capillaroscopy system. The measurements of the velocity were performed at 12 points in nailfold regions by two observers separately. The image processing and automated measurement take 1-2 min per nailfold. FFmpeg software was used to convert the images format, and then the images were imported to ImageJ software and segmented. SIFT tracking algorithms and Kalman filter were used to filter noise and irregularities in the images. For reproducibility, the velocity distribution values obtained by the two performers, and Paired T-Test was used. The reliability of a measuring instrument or calculation method depends on the tools obtained using Cronbach's alpha. To assess the repeatability of the algorithm, the capillary velocity values were executed at different times with 24-hour intervals using a coefficient of variance method. Results: The calculated RBCs velocity was in the range of 0.05-0.16 mm/s. The results based on Cronbach's alpha analysis for reliability factor was 0.97, with a good correlation among the measurements, 0.85. The average velocity (along with standard deviation) for repeatability at three different times was obtained 0.1195 ± 0.0246, 0.0974 ± 0.0221, and 0.0962 ± 0.0202 mm/s, demonstrating that there was no statistically variation between these measurements (P-value > 0.05). The velocity results for the two observers were 0.811 ± 0.392 and 0.819 ± 0.325 mm/s, indicating a good reproducibility between them (P-value = 0.959). Conclusion: For the measurements of nailfold capillaries velocity, there was good/reasonable reliability, repeatability, and reproducibility. The results indicated a good accuracy of capillaroscopy system and ImageJ software with SIFT algorithm and Kalman filter, which can be used as an appropriate tool for determining the rate of nailfold blood flow velocity.