This study presents a new methodology that combines quantitative image analysis, clustering, and statistical techniques to examine the 2D distribution of osteohistological features in an extinct stem-group bird. The driving force of our research was the need to map ontogenetic changes in the spatial density of osteocyte lacunae in the type specimen of Musivavis amabilis from the Lower Cretaceous of China. We particularly focused on developing tools to reveal quantitative aspects of the dynamics in the formation of avascular bone in this active flier. We achieve this goal by proposing an algorithm with the following methodological steps: 1) We obtain relevant coordinate details for pixel locations selected by thresholding the original image with shading as a criterion. 2) We estimate density using the Gaussian kernel estimator and refined it through observations and regression analysis. 3) After slicing the image, we apply k-means clustering to obtain one-dimensional representations of lacunar density. 4) We proceed by implementing weighted averaging employing the k-nearest neighbor approach. Having applied these steps, we are able to quantitatively disclose growth processes previously unnoticed and reveal dynamics in the formation of lacunar bone tissue in the enantiornithine birds capable of power flight.
Read full abstract