Automatic atlas-based contour extraction of anatomical structures in medical images

We present an extension of the live-wire approach to an automatic segmentation method. Similarities between anatomical structures in different patient data are exploited by transferring structure outlines from atlas to patient data. After preprocessing via affine registration methods, the automatic segmentation is done slice by slice. Atlas seed points are relocated in each patient slice and then connected by cost-trained live-wires. After the automatic contour extraction, an efficient graph search algorithm facilitates the interactive adjustment of the contours created. The evaluation with hip CT data showed that the tool simplifies the segmentation of anatomical hip structures because 51% of interaction time could be saved maintaining the segmentation quality compared to manually created segmentations.