Application of support vector clustering to the visualization of medical images

A support vector machine (SVM) based method, support vector clustering, is applied to the problem of modelling 3D objects represented in CT medical images. The method produces accurate surface representations of the objects from data distributed in its volume. This procedure is of advantage in medical imaging since it does not require complicated segmentations and it is shown to be noise robust. There seems to be no limitations regarding the topology of the object to be modelled and a high number of data points can be processed. The method outputs sparse results in the sense that the model is defined in terms of a significant reduced set of data points, achieving great compression rates.