Purchase this article with an account.
Che Liu, Hong Jiang, Jianhua Wang, Zhe Xu, Aizhu Tao, Delia DeBuc; Automated quantitative analysis of conjunctival microcirculation. Invest. Ophthalmol. Vis. Sci. 2014;55(13):2775.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
To demonstrate automatic image processing to quantitatively analyze conjunctival microcirculation.
Video sequences with a region of interest (ROI, 2.82 × 2.10 mm2) of the bulbar conjunctiva were acquired using our newly developed functional slit-lamp biomicroscope (FSLB) with a high magnification (210x) and a high speed (60 frames/sec). To obtain conjunctiva blood vessel diameter, blood velocity and flow rate, custom software was developed. The processed videos were in audio video interactive (AVI) file format, which were converted to consecutive frames. In order to compensate eye motion during image acquisition, automatic image registration was performed based on correlation coefficients. After registration, a series of image processing steps were performed including averaging frames to reduce noise, time-averaged image enhancement and vessel segmentation and morphological operations. Then vessels were detected automatically to calculate the length and diameter of each vessel. Axial red blood cell (RBC) velocity was estimated using a spatial-temporal image (STI) composed of the location of RBCs as a function of the image frame. An automatic procedure was implemented for the blood flow velocity calculation by applying the Hough transform. Visual inspection was conducted and necessary correction was done during processing STI images (Fig. 1). Furthermore, multiple ROIs were registered to create a wide field (Fig.2). Six video sequences with different ROIs of one eye were analyzed.
According to visual inspection, more than 80% of vessels can be automatically detected and segmented from the time-averaged image of frames. The STI revealed distinct bands, each corresponding to either the movement of RBCs or the space between them in each image frame. The mean slope of the identified vessel band, which is a measure of axial RBC velocity, fitted well the longest continuous band of RBC movements in the STI. The software implemented allows the user to trace lines in the STI interactively, when the automatic velocity calculation fails. From six different ROIs, the mean vessel diameter was 27.2 ± 5.1 µm. The blood flow velocity was 0.84 ± 0.29 mm/s and flow rate was 350.1 ± 189.2 pl/s. A wide field (Fig 2) was successfully reconstructed from these 6 small ROIs.
We demonstrated the feasibility of automated image processing of conjunctival microcirculation.
This PDF is available to Subscribers Only