Kernel SIMPLE: Positron Lifetime Image Reconstruction Using Kernel Method

Ortho-positronium (o-Ps) lifetime has the potential to become a biomarker for quantifying oxygen concentration and micro-structural properties of human tissue in vivo. Several model-based lifetime image reconstruction methods have been developed. The most computationally efficient method is the Statistical Image Reconstruction of Positron Lifetime via Time-Weighting (SIMPLE) method. It can achieve a spatial resolution comparable to that of the standard positron emission tomography (PET) images with a computational cost equivalent to two standard PET image reconstructions. However, the method lacks effective regularization to suppress noise in the lifetime images. To address this issue, here we developed a kernel-based SIMPLE method. The kernel matrix is first constructed using either dynamic activity images or lifetime-thresholded images and then utilized to represent both the activity image and the lifetime-weighted image in the SIMPLE method. Computer simulations show the kernel SIMPLE method can substantially reduce noise in lifetime images compared to the original SIMPLE method. The dynamic-image-based kernel matrix demonstrates better noise-suppression capability compared to the lifetime-thresholded kernel matrix.