ABSTRACT Glioma make up 80% of all primary malignant brain tumors. The current standard treatment for newly diagnosed gliomas includes maximal surgical resection, radiation therapy (RT), and chemotherapy. A key technical challenge in RT treatment planning is accurate target volume delineation of gliomas. The current clinical guidelines for target volume delineation rely primarily on structural Magnetic Resonance Imaging (MRI) images. Gross Tumor Volume (GTV) is defined based on contrast-enhanced T1-weighted MRI and T2-weighted MRI. However, structural MRI alone lacks specificity for delineation of true tumor boundaries. Accordingly, Clinical Target Volume (CTV) is often defined as the GTV plus a large margin (e.g., 20-25 mm) to account for possible microscopic infiltration. The lack of specificity of structural MRI is a critical factor limiting the investigation and clinical application of new RT techniques for better clinical outcome. MR spectroscopic imaging (MRSI) has long been recognized as a potentially powerful tool for label-free molecular imaging of brain tumor. In a recent Phase I clinical trial, MRSI is used to guide dose escalation in RT for Glioblastoma multiforme patients, showing very promising preliminary results. Although general clinical applications of MRSI have been impeded by its limited spatial resolution and long scan time, significant progresses have been made in addressing these technical challenges over the past decade using advanced data acquisition and processing methods. Our group have successfully developed a powerful MRSI technology, known as SPICE (SPectroscopic Imaging by exploiting spatiospectral CorrElation). SPICE effectively integrates rapid scanning, sparse sampling, quantum simulation of molecule resonance structures, and machine learning to enable rapid high-resolution MRSI. Preliminary results by our and other groups have shown an exciting potential of SPICE to achieve an unprecedented combination of resolution, speed, and SNR for metabolic imaging. We have also demonstrated, for the first time, the feasibility of mapping T1, T2 and proton-density parameters of brain tissues using the unsuppressed water signals from the SPICE scans. The primary goal of this project is to leverage this significant advance in MRSI technology and investigate the use of high-resolution metabolic and structural information to achieve more accurate target volume delineation for RT treatment planning of gliomas. We will: 1) further develop and optimize SPICE for MRI/MRSI-guided RT of gliomas in clinical settings, 2) perform systematic performance evaluation of the proposed method on phantoms, healthy subjects, and glioma patients, and 3) investigate the use of metabolic and structural biomarkers for delineation of biological target volume to improve image-guided RT of gliomas. The proposed research is innovative in developing a novel molecular imaging technology and a timely effort on improving RT treatment planning of gliomas with quantitative metabolic and structural biomarkers. Successful completion of the project will have a significant impact on image-guided RT for gliomas, opening up new opportunities for better control of recurrence in glioma patients using dose escalated RT.

NARRATIVE Glioma is the most commonly occurring type of malignant brain tumor in the United States. This project will improve image-guided radiation therapy of gliomas by using both metabolic and structural biomarkers obtained from a new magnetic resonance spectroscopy imaging technology for accurate delineation of tumor boundaries.