Scale Space-based Capillary Detection to Determine Capillary Density and Avascular in Nailfold Capillary Images Using USB Digital Microscope
Nailfold capillaroscopy (NC) is a non-invasive technique used for detecting multiple medical disorders such as connective tissue, rheumatic and systemic diseases. Current NC methods involve acquiring high-resolution images from expensive video capillaroscope for analysis. NC analysis on low-resolution images obtained from a low-cost hardware is a challenging task. Scale space capillary detectors (SCD) is proposed for detection of capillaries under such conditions, which is a unique combination of anisotropic diffusion and Harris corner detector. SCD is followed by Ordinate clustering algorithm (OCA) which is used to eliminate outliers in the detected capillaries. Nailfold capillary images are obtained from a low-cost digital microscope under varying lighting conditions to form a custom database. Experimental results show the promising performance of the proposed algorithm with a high true-positive detection rate and a low false-negative detection rate.
KeywordsNailfold capillary Anisotropic diffusion Ordinate clustering algorithm Bi-cubic interpolation Harris corner detector
- 3.Harris, C., Stephens, M.: A combined corner and edge detector. In: Proceedings of Alvey Vision Conference, vol. 15, pp. 147–151 (1988)Google Scholar
- 7.Srivastava, A., Bhateja, V., Tiwari, H.: Modified anisotropic diffusion filtering algorithm for MRI. In: Proceedings of (IEEE) 2nd International Conference on Computing for Sustainable Global Development 1885–1890 (2015)Google Scholar