For the alignment quantification, in particular for fiber-based or ROI-based ones, the number of fiber or fiber segments should be larger than ten to yield a statistically robust measurement. Although the alignment coefficient and mean of orientations or angles are the main metrics used for quantifying fibrillar collagen alignment, CurveAlign also outputs other orientation-related metrics such as the variance, standard deviation, median, skewness, Kurtosis, etc. to characterize the distribution or dispersion of collagen fiber orientations. These provide additional useful information for some potential applications.
Visual inspection of the output images, which includes extracted fibers or curvelets overlaid on the original image, enables the user to assess the results for any image without the need for comparing against a ground truth. In addition, we can compare the statistics change by adjusting the primary parameters mentioned above. In many cases we tested, the alignment or orientation did not have significant change if the visual inspection is properly conducted.
Visual inspection of the output images, which includes extracted fibers or curvelets overlaid on the original image, enables the user to assess the results for any image without the need for comparing against a ground truth. In addition, we can compare the statistics change by adjusting the primary parameters mentioned above. In many cases we tested, the alignment or orientation did not have significant change if the visual inspection is properly conducted.
