Nuclear medicine imaging in children has been shown to have significant clinical value across all organ systems. In providing this significant benefit it is critical to minimize the radiation dose used in pediatric patients, whose risk for adverse health effects (such as cancer) per unit administered activity is much higher than that of adults, owing to their higher tissue sensitivity and longer potential lifespan. The governing principle of this project will be to minimize radiation dose while methodically ensuring that lesion detection performance is fully preserved. This will be accomplished by using validations based on both numerical and physician observers measuring performance in tasks that emulate those performed clinically. We will employ two approaches in tandem to enable lowering dose while maintaining performance. First, we will use advanced image reconstruction and processing techniques. Corrections for various forms of image quality degradation will be incorporated in the reconstruction, and deep learning (DL) will be used for post-reconstruction denoising. Second, we will develop methods to correct for both body and respiratory motion, which degrade diagnostic accuracy. Correcting for body and respiratory motion will allow dose to be reduced without loss of image quality and will also offer a technological alternative to using sedation or even general anesthesia to minimize motion when imaging children. For this investigation we have selected 99mTc-labeled dimercaptosuccinic acid (DMSA) renal imaging as a testbed to demonstrate our approaches. Damage to the renal cortex resulting from infection of the kidneys is a critical issue in children, including newborns and toddlers. DMSA SPECT is the ?gold-standard? in the evaluation of pyelonephritis and renal scarring post- infection. The concepts we will demonstrate for reduction of radiation dose and correction of motion with DMSA will be translatable to other SPECT (and PET) studies in pediatric imaging and beyond.
Our Specific Aims are: 1. Establish infrastructure for investigating and evaluating advanced reconstruction and motion correction; 2. Determine the extent of radiation dose reduction to pediatric patients through improved reconstruction and DL denoising while maintaining optimal full-dose lesion detection accuracy; 3. Develop data-driven and depth-sensing camera methods for body and respiratory motion estimation and correction; and 4. Conduct numerical and physician observer studies to validate the level of dose reduction enabled by DL denoising and motion correction.
In nuclear medicine imaging, it is critical to minimize the radiation dose used in pediatric patients, whose risk for adverse health effects (such as cancer) per unit administered activity is much higher than that in adults, owing to their higher tissue sensitivity and longer potential lifespan. Correcting for body and respiratory motion occurring during imaging will improve the quality of the formed three-dimensional images of the patient by reducing blurring and image artifacts and offer a technological alternative to using sedation or even general anesthesia to reduce motion when imaging children, which can bear health risks of its own. We propose an advanced reconstruction methodology which would enable reduction in the amount of activity administered and compensate for patient motion during imaging.
