Method and apparatus for learning-enhanced atlas-based auto-segmentation

摘要

Disclosed herein are techniques for enhancing the accuracy of atlas-based auto-segmentation (ABAS) using an automated structure classifier that was trained using a machine learning algorithm. Also disclosed is a technique for training the automated structure classifier using atlas data applied to the machine learning algorithm.

主权项

1. A method for training a classifier, the method comprising:
receiving a predetermined threshold distance; selecting, with a processor, a plurality of training samples from an atlas image having at least one pre-identified structure of interest, wherein the atlas image includes a plurality of image data points, wherein the plurality of training samples are randomly selected from image data points located within the predetermined threshold distance from a contour of the structure of interest; determining, with the processor, a set of image attributes associated with each selected training sample; and applying, with the processor, the selected training samples and the image attributes associated with the selected training samples to a machine-learning algorithm to generate a structure classifier, the structure classifier being configured to determine a structure of interest in a subject image.