Multi-Atlas Segmentation Using Partially Annotated Data: Methods and Annotation Strategies.

Area of Science:

• Medical Image Analysis
• Computer Vision
• Computational Anatomy

Background:

• Multi-atlas segmentation leverages annotated datasets for robust medical image analysis.
• Limited availability of fully annotated atlases due to extensive labeling time is a major bottleneck.
• Existing methods often require complete atlas annotations, increasing expert workload.

Purpose of the Study:

• To develop a unified framework for multi-atlas segmentation.
• To enable segmentation using partially annotated atlas images, reducing annotation burden.
• To investigate the efficacy of sparse annotation strategies in multi-atlas segmentation.

Main Methods:

• Re-formulated the labeling problem as a Markov Random Field energy minimization on a graph.
• Modified graph configurations to accommodate partially annotated atlas images.
• Evaluated on Magnetic Resonance Imaging (MRI) datasets for hippocampal and cardiac segmentation.

Main Results:

• The proposed framework successfully recreated existing multi-atlas segmentation techniques.
• Demonstrated the potential of using sparsely annotated atlas data for accurate segmentation.
• Partial annotation strategies were investigated and validated.

Conclusions:

• The proposed unified framework effectively handles multi-atlas segmentation.
• Partially annotated atlas data can significantly reduce expert annotation effort without compromising accuracy.
• This approach offers a more efficient solution for medical image segmentation.