Abstract
OBJECTIVES
Neurosurgical resection in gliomas depends on the precise preoperative definition of the tumor and its margins to realize a safe maximum resection that translates into a better patient outcome. New metabolic imaging techniques could improve this delineation as well as designate targets for biopsies. We validated the performance of our fast high-resolution whole-brain 3D-magnetic resonance spectroscopic imaging (MRSI) method at 7T in high-grade gliomas (HGGs) as first step to this regard.
METHODS
We measured 23 patients with HGGs at 7T with MRSI covering the whole cerebrum with 3.4mm isotropic resolution in 15 min. Quantification used a basis-set of 17 neurochemical components. They were evaluated for their reliability/quality and compared to neuroradiologically segmented tumor regions-of-interest (necrosis, contrast-enhanced, non-contrast-enhanced+edema, peritumoral) and histopathology (e.g., grade, IDH-status).
RESULTS
We found 18/23 measurements to be usable and ten neurochemicals quantified with acceptable quality. The most common denominators were increases of glutamine, glycine, and total choline as well as decreases of N-acetyl-aspartate and total creatine over most tumor regions. Other metabolites like taurine and serine showed mixed behavior. We further found that heterogeneity in the metabolic images often continued into the peritumoral region. While 2-hydroxy-glutarate could not be satisfyingly quantified, we found a tendency for a decrease of glutamate in IDH1-mutant HGGs.
DISCUSSION
Our findings corresponded well to clinical tumor segmentation but were more heterogeneous and often extended into the peritumoral region. Our results corresponded to previous knowledge, but with previously not feasible resolution. Apart from glycine/glutamine and their role in glioma progression, more research on the connection of glutamate and others to specific mutations is necessary. The addition of low-grade gliomas and statistical ROI analysis in a larger cohort will be the next important steps to define the benefits of our 7T MRSI approach for the definition of spatial metabolic tumor profiles.
Ensemble segmentation using AdaBoost with application to liver lesion extraction from a CT volume
