Reconstruction/segmentation of attenuation map in TOF-PET based on mixture models

Area of Science:

• Medical Imaging
• Nuclear Medicine
• Image Reconstruction

Background:

• Accurate attenuation correction is crucial for quantitative imaging in Positron Emission Tomography (PET).
• Emission-based methods offer a promising approach for attenuation map estimation in Time-of-Flight PET (TOF-PET) scanners.
• Existing methods may face challenges with noise and spatial uniformity in reconstructed attenuation maps.

Purpose of the Study:

• To develop an enhanced emission-based method for improved attenuation map estimation in PET.
• To integrate histogram-based information and spatial priors to refine attenuation correction accuracy.
• To enable simultaneous segmentation of the attenuation map during reconstruction.

Main Methods:

• A maximum a posteriori (MAP) framework was employed, incorporating a mixture model prior with histogram-based information.
• A Median root prior was integrated to reduce inter-voxel noise and promote spatial smoothness.
• Joint-MAP optimization was performed iteratively, alternating between activity and attenuation updates with histogram mixture decomposition.

Main Results:

• The proposed method successfully generated nearly segmented attenuation maps.
• Incorporation of histogram priors and Median root prior improved the performance of the emission-based approach.
• The method demonstrated effective noise reduction and enhanced spatial smoothness in reconstructed attenuation maps.

Conclusions:

• The developed method shows significant potential for accurate attenuation correction in PET imaging.
• The approach can be utilized as a standalone technique or in conjunction with other methods for PET/MR systems.
• The simultaneous segmentation capability offers an advantage for attenuation map reconstruction.