https://www.selleckchem.com/products/CHIR-258.html Interpreting ultrasound (US) images of the spine is challenging due to the high variability of the contrast during freehand US acquisitions. In this paper, an automatic method to extract vertebral landmarks (spinous process and laminae) from US images acquired in the transverse plane is presented. Prior knowledge about the vertebral shape and the associated hyper-echoic property is incorporated using the horizontal and vertical projections of the image intensities. After detrending, the mean-value crossing of the projections is used to define the concept of mean boundary and locate landmarks without the need for thresholding or parameter adjustment. The method was evaluated using two datasets a porcine cadaver dataset (PC) with CT data registered to the US data used as a gold standard, and a healthy human subjects dataset (HH) with a silver standard generated from manual landmarks located on the US data acquired with a curvilinear (6C2) and linear (14L5) probe. The mean sum of distances (MSD) of the landmark extraction to the gold and silver standards is respectively MSD=0.90±1.05 mm for PC, MSD=1.14±1.08 mm (6C2) and MSD=3.54±2.69 mm (14L5) for HH. Results are satisfying on PC and HH with 6C2. Variable contrast quality for 14L5 gives satisfying results for the spinous process but not for the laminae. The proposed approach has the potential to be used for different applications in the context of US spine imaging such as scoliosis follow-up and intra-operative surgical guidance.The calculation of the largest Lyapunov exponent (LyE) requires the reconstruction of the time series in an N-dimensional state space. For this, the time delay (Tau) and embedding dimension (EmD) are estimated using the Average Mutual Information and False Nearest Neighbor algorithms. However, the estimation of these variables (LyE, Tau, EmD) could be compromised by prior filtering of the time series evaluated. Therefore, we investigated the e