1: \begin{abstract}

2: 	

3: %% Text of bstract

4: %\textit{Shape} is all the geometrical information that remains when location, scale and rotational effects are removed from an object.

5: %Modelling shape data has recently gained interest in many research fields

6: %since any proper shape inference should rely on the geometric structure of the objects

7: %under study. Relevant applied fields are e.g. biology, medical imaging, computational anatomy, computer vision, archaeology and chemistry.

8: %From a theoretical point of view, the involved data spaces are not Euclidean hence differential geometric tools should be adopted in order to develop a proper statistical shape theory. Some descriptive and basic inference

9: %tools have been proposed in literature but to the current knowledge no regression methodologies have been proposed a part from \cite{dryden1},

10: %who propose a marginal likelihood approach to implement a size-and-shape

11: %response regression with Gaussian landmarks.

12: %Likelihood-based approaches do not efficiently work with complicated covariance structures and moreover could be compromised by convergence issues of numerical optimization algorithms. An alternative easily-interpretable

13: %approach which fits into the framework of Bayesian latent variable models

14: %potentially allows the specification of more complicated covariance structures

15: %and only requires a posterior approximation by means of MCMC methods.

16: Building on \cite{dryden1}, this note presents the Bayesian estimation of a regression model for size-and-shape response variables with Gaussian landmarks. Our proposal fits into the framework of Bayesian latent variable models and allows a highly flexible modelling framework.% which is still missing in the statistical literature. %\textit{Shape} is all the geometrical information that remains when location, scale and rotational effects are removed from an object.

17: 

18: \end{abstract}

19: