AI-Enhanced Perfusion MRI for Optimizing Y90 Particle Density in Radioembolization of Liver Cancer

About This Grant

PROJECT SUMMARY Radioembolization (RE) with yttrium-90 (90Y) microspheres is a promising treatment for primary and secondary liver cancer, but the success of the treatment is user-dependent and relies on the optimal distribution of 90Y particles within the tumor. The current method for determining how 90Y microspheres are dispersed in a tumor requires the tracking of Tc-99m-labeled macro-aggregated albumin (MAA) distribution after the mapping procedure. This methodology is labor-intensive and cumbersome, and does not account for flow dynamics within the tumor. Variations in peri-tumoral and intra-tumoral blood flow can significantly impact final 90Y microsphere distribution, leading to undertreated cold spots or overtreated embolic regions that cause beads to reflux backwards. There is thus an unmet need for an advanced tool to predict 90Y distribution accurately based on tumor flow characteristics. This project aims to overcome this current barrier by developing an in vivo imaging biomarker of 90Y microsphere distribution that leverages AI-optimized perfusion analyses and a biological model of liver flow dynamics. The project will be conducted via two aims: 1) Development of an AI-optimized liver perfusion analysis tool using routine clinical MRI scans. Quantitative dynamic contrast-enhanced (DCE) MRI will be used to obtain insights into tumor blood flow and vascularity. With standardized data collection and curation, we will develop advanced deep learning-enabled perfusion analysis to analyze liver perfusion characteristics in routine clinical DCE-MRI. A physical model of particle flow will be integrated with perfusion MRI for accurate prediction of RE microsphere distribution, and preclinical evaluation will be conducted with patient- derived liver tumor phantoms, allowing for precision measurements of the MRI-based perfusion protocol. 2) In vivo validation of AI-optimized perfusion protocol for accurate prediction of 90Y particle densities. We will test intra-subject and inter-subject variability of the perfusion MRI protocol by repeating DCE-MRI scans at multiple time points with concomitant repositioning. Once intra-subject and inter-subject reproducibility are validated, the AI-optimized perfusion MRI will be leveraged to quantify perfusion parameters in an in vivo porcine liver cancer model intra-arterially infused with 90Y microspheres. Histopathological and imaging analyses of 90Y microsphere distribution will then be correlated with the MRI perfusion markers against those from the standard Tc-99m MAA administration. The results of this project will have significant implications for the treatment of liver tumors by providing a foundation for more effective and precise 90Y-based therapies. By optimizing 90Y particle density with tumor perfusion characteristics, we can improve the outcomes and safety of RE, ultimately transforming it into a more effective and precise therapeutic approach for liver cancer patients.