Dual-Layer Groupwise Registration for Consistent Labeling of Longitudinal Brain Images

Area of Science:

• Neuroimaging
• Medical Image Analysis
• Computational Neuroscience

Background:

• Longitudinal brain imaging is crucial for disease diagnosis, requiring robust registration and labeling methods.
• Existing methods often neglect temporal consistency and the manifold structure of longitudinal image data.
• Independent atlas alignment to each time point limits accuracy and consistency in longitudinal studies.

Purpose of the Study:

• To develop an advanced dual-layer groupwise registration method for consistent anatomical labeling of longitudinal brain images.
• To enhance the accuracy and temporal consistency of brain image labeling in disease diagnosis.
• To address the limitations of existing methods in handling the unique characteristics of longitudinal image data.

Main Methods:

• A novel dual-layer groupwise registration framework is proposed, utilizing a multi-atlas labeling approach.
• Atlases are jointly aligned to all time-point images via a subject-mean image, acting as a bridge.
• Inter-atlas relationships within their manifold guide the registration of each atlas to the subject-mean image.

Main Results:

• The dual-layer method achieved higher labeling accuracy and maintained temporal consistency compared to traditional registration schemes.
• Experiments on healthy infant and Alzheimer's disease datasets demonstrated the method's effectiveness.
• The framework showed improved labeling accuracy when integrated with existing label fusion techniques.

Conclusions:

• The proposed dual-layer groupwise registration method significantly enhances the consistency and accuracy of anatomical labeling for longitudinal brain images.
• This approach offers a more robust solution for analyzing brain changes over time in disease progression studies.
• The method provides a flexible framework adaptable to various label fusion techniques for improved diagnostic tools.