Anatomy segmentation through low-resolution multi-atlas label fusion and corrective learning

摘要

Computationally efficient anatomy segmentation through low-resolution multi-atlas label fusion and corrective learning is provided. In some embodiments, an input image is read. The input image has a first resolution. The input image is downsampled to a second resolution lower than the first resolution. The downsampled image is segmented into a plurality of labeled anatomical segments. Error correction is applied to the segmented image to generate an output image. The output image has the first resolution.