Research Project
9680585
Preclinical positron emission tomography (PET) plays a significant role in monitoring radiotracer biodistributions for the development of new therapies. The standardized uptake value (SUV) has become one of the most important quantitation measures in PET analysis. However, SUV extraction from PET data hinges on the ability to clearly delineate the organ region-of-interest (ROI) from an anatomical reference. Cross-comparison of different organ SUVs and biodistributions is also cumbersome due to complex registration strategies of each animal with an unknown pose and size. Operator bias and lack of efficient co-registration strategies results in poor data reproducibility, high data variability, low data throughput and prohibits the use of fully-automated data parsing and data analysis for predicting early therapy outcomes with high sensitivity. In Vivo Analytics will directly address these shortcomings by developing InVivoPET. It will be a cloud-based PET data analysis tool, which will enable automatic organ SUV extraction followed by an instantaneous biodistribution analysis. InVivoPET will automatically coregister PET images to the animal?s anatomy and will calculate biodistributions in almost real-time. InVivoPET consists of several parts. First, a Body Conforming Animal Mold (BCAM) enables consistent spatial and longitudinal registration of the animal?s pose and location to the PET data. Second, a statistical mouse atlas based on an Organ Probability Map (OPM) provides a digital and operator-independent organ ROI template. Third, a cloud-based software with a browser-based user interface enables an automatic organ SUV extraction with following biodistribution analysis. Last, machine learning and data mining algorithms can be applied in the future that will further enhance study outcomes. InVivoPET does not rely on manual delineation of organs. A machine-driven data analysis fully eliminates operator-dependent variability and increases data reproducibility. It will enable the drug development team to quantitate the impact of candidate therapeutics with the highest accuracy, reduces the time to enter clinical trials, reduces costs, and ensures the quantification and consistency of PET data. Therefore, the hypothesis is that the organ SUV can automatically be extracted from PET data using the BCAM and OPM. In Aim 1, we will modify the BCAM for housing mice with surface-protruding tumors and confirm the ability for spatially aligning mice with tumor xenografts to the OPM. In Aim 2, we will perform PET imaging of tumor bearing mice using 18F-FDG and confirm the ability for automatically extracting the organ SUV based on the OPM. The successful completion of the proposed project will help to commercialize InVivoPET, which will be sold as a plug-in to existing PET systems and as a PET- manufacturer independent Software-as-a-Service (SaaS).
