Full-stack automation for reliable and reproducible MRS of brain cancer - PROJECT SUMMARY/ABSTRACT
This project proposes to develop methods for automated, real-time, single-voxel magnetic resonance
spectroscopy (MRS) in brain tumors, integrate these methods with a clinical MRI system, evaluate their
performance, and make them available as open-source tools to the research community. MRS can provide
metabolic information noninvasively for assessment of tumor phenotype and therapeutic response. Single-
voxel MRS methods provide the best quality and most reliable data, but require the scanner operator to have a
high skill level and expertise to produce good quality results. The need for this expert involvement in both
acquisition and analysis remains a critical barrier to the translation of MRS methods to clinical research sites
without spectroscopy experts and to clinical practice. The first part of this project is to develop a method for 3D
voxel placement using image guidance, integrate this method with a clinical MR system, and evaluate its
performance. In the second part, we will automate our advanced MRS methods. In the third part, we will create
real-time, automatic quantification tool specific to the obtained MRS data that will provide clinically interpretable
results. In the final part, we will assess the performance of automated methods in prospective in vivo study.
Successful completion of this project will improve data robustness and quality, eliminating the need for the
expert interaction at the time of the scan and enabling adoption of MRS in multi-site clinical trials and clinical
practice.