Preliminary testing of sensitivity to input data quality in an elastographic reconstruction method

An elastographic reconstruction method has been developed to recover the material properties of soft tissue by model-based analysis of image data acquired at different states of mechanical loading. The algorithm utilizes image similarity as part of the cost function for a multi-resolution, non-linear optimization. Previous work with a phantom membrane used for simulated dermoscopic application has prompted this preliminary investigation of the relative effects of additive image noise and boundary condition determination errors on the performance of the method. The results as quantified by elasticity contrast and localization accuracy indicate that the reconstruction process is robust in the presence of realistic levels of image corruption and tolerates the majority of boundary condition mapping errors