Medical image segmentation and atlas image selection

Abstract

Some embodiments relate to segmentation of medical images. For example, a medical image may be registered to multiple atlas images, after which a segmentation function may be applied. Multiple partitions may be fused into a final overall partition. The atlas image may be selected based on a high segmentation quality or a low registration quality.

Description

technical field [0001] The presently disclosed subject matter relates to methods for medical image segmentation, methods for selecting atlas images for use in medical image segmentation methods, medical image segmentation systems, atlas image selection systems, and computer-readable media. Background technique [0002] Accurate and robust medical image segmentation remains a challenging task. [0003] One approach to medical image segmentation is a multi-atlas-based image segmentation approach. These methods can be built with a limited set of segmented reference images, called an atlas. Unfortunately, atlas-based methods often lack the required accuracy. The registration method and its methods are given in "Deformable medical image registration: a survey" by A. Sotiras et al. or in "Symmetric diffeomorphic image registration with cross-correlation: evaluating automated labeling of elderly and neurodegenerativebrain" by B.B. Avants et al. Examples of applications to medica...

AI Technical Summary

Problems solved by technology

Unfortunately, neural networks do not generalize properly to new datasets with slightly different properties

When a trained neural network is presented with a new image, even if the new image is slightly outside the image used to train the neural network, the resulting segmentation may be completely incorrect

Images