Abstract Total-body (TB) Positron Emission Tomography (PET) is a medical imaging technology that enables comprehensive imaging of metabolic processes across the entire body in a single bed position. It offers substantial advantages for dynamic imaging and kinetic modeling, such as high temporal resolution and multiparametric capabilities, making it invaluable for assessing cancer staging, early treatment response, and distinguishing malignancy from infection/inflammation. However, the widespread adoption of TB PET is limited by its high cost, primarily due to the expense of scintillator material and readout electronics. This proposal seeks to develop an affordable dynamic TB PET system through the use of sparse detector architectures, which will reduce material costs while maintaining the performance benefits of full-body coverage. We hypothesize that sparse designs can produce high-quality dynamic images that allow for kinetic modeling and parametric imaging with minimal loss of quantitative accuracy compared to full-rank TB PET. Our innovative approach combines spatial down sampling of the detector matrix of the 2-m long EXPLORER TB PET scanner with advanced reconstruction algorithms. Classical and deep learning-based Kernel methods will help recover image quality in high-noise, low-count environments. This project will address three specific aims: (1) demonstrating parametric imaging using sparse TB PET, (2) comparing image quality between sparse TB PET and conventional PET scanners, and (3) optimizing reconstruction parameters and detector matrix configurations for high temporal resolution in dynamic imaging. By achieving these aims, we will develop a robust, cost- effective TB PET system with potential to revolutionize both clinical and research applications. The successful implementation of this technology will expand access to dynamic TB PET, enhance diagnostic workflows, and accelerate research in cancer and other diseases.