Addressing Partial Volume Effects in Clinical PET Quantification: Modern Correction Strategies and Challenges

Area of Science:

• Medical Imaging
• Nuclear Medicine
• Image Processing

Background:

• Accurate quantification in clinical positron emission tomography (PET) is crucial for diagnosis and treatment monitoring.
• Partial volume effects (PVEs) in PET imaging lead to underestimation of tracer uptake in small lesions and organs.
• Existing partial volume correction (PVC) methods aim to mitigate these PVEs for improved quantitative accuracy.

Purpose of the Study:

• To categorize and review novel partial volume correction (PVC) methods for clinical positron emission tomography (PET).
• To provide an overview of the different approaches available for improving quantitative accuracy in PET imaging.

Main Methods:

• Categorization of novel PVC methods into four main groups: post-reconstruction region-based, post-reconstruction image enhancement, in-reconstruction resolution modeling, and deep learning-based approaches.
• Description of the underlying principles and characteristics of each category.
• Highlighting the suitability of each method for quantification versus visual analysis.

Main Results:

• Post-reconstruction region-based methods provide regional values suitable for quantification but not visual analysis.
• Image enhancement techniques utilize deconvolution or wavelet domain combinations for improved image quality.
• In-reconstruction methods incorporate resolution modeling within iterative algorithms.
• Deep learning approaches offer versatile training strategies using simulated, real, or self-supervised data.

Conclusions:

• Novel PVC methods offer diverse strategies to address partial volume effects in PET imaging.
• The choice of PVC method depends on the specific application, balancing quantitative accuracy and visual interpretability.
• Deep learning-based methods represent a promising frontier in developing advanced PVC techniques.