Research Articles

Download PDF (20.66 MB)

TOMOGRAPHY, June 2016, Volume 2, Issue 2: 106-116
DOI: 10.18383/j.tom.2016.00136

A Systematic Pipeline for the Objective Comparison of Whole-brain Spectroscopic MRI with Histology in Biopsy Specimens from Grade III Glioma

J. Scott Cordova1, Saumya S. Gurbani1,2, Jeffrey J. Olson3,4, Zhongxing Liang1, Lee A. D. Cooper2,5, Hui-Kuo G. Shu4,6, Eduard Schreibmann6, Stewart G. Neill7, Constantinos G. Hadjipanayis3,8, Chad A. Holder1, and Hyunsuk Shim1,2,4

1Department of Radiology and Imaging Sciences, Emory University School of Medicine, Atlanta, Georgia; 2Department of Biomedical Engineering, Georgia Institute of Technology, Atlanta, Georgia; 3Department of Neurosurgery, Emory University School of Medicine, Atlanta, Georgia; 4Winship Cancer Institute of Emory University, Emory University, Atlanta, Georgia; 5Department of Biomedical Informatics, Emory University School of Medicine, Atlanta, Georgia; 6Department of Radiation Oncology, Emory University School of Medicine, Atlanta, Georgia; 7Department of Pathology, Emory University School of Medicine, Atlanta, Georgia; and 8Department of Neurosurgery, Icahn School of Medicine at Mount Sinai, New York, New York


The diagnosis, prognosis, and management of patients with gliomas are largely dictated by the pathological analysis of tissue biopsied from a selected region within the lesion. However, the heterogeneous and infiltrative nature of gliomas make it difficult to identify the optimal region for biopsy with conventional magnetic resonance imaging (MRI). This is particularly true for low-grade gliomas, which are often nonenhancing tumors. To improve the management of patients with such tumors, neuro-oncology requires an imaging modality that can specifically identify a tumor’s most anaplastic/aggressive region(s) for biopsy targeting. The addition of metabolic mapping using spectroscopic MRI (sMRI) to supplement conventional MRI could improve biopsy targeting and, ultimately, diagnostic accuracy. Here, we describe a pipeline for the integration of state-of-the-art, high-resolution, whole-brain 3-dimensional sMRI maps into a stereotactic neuronavigation system for guiding biopsies in gliomas with nonenhancing components. We also outline a machine-learning method for automated histological analysis that generates normalized, quantitative metrics describing tumor infiltration in immunohistochemically stained tissue specimens. As a proof of concept, we describe the combination of these 2 techniques in a small cohort of patients with grade 3 glioma. With this work, we aim to present a systematic pipeline to stimulate histopathological image validation of advanced MRI techniques, such as sMRI.

Supplemental Media

  • Video 1: Image analysis pipeline for SOX2 density calculation. Initial model estimates were calculated based on simple peak detection in the pixel intensities of the hematoxylin channel in this Supplementary video. Least-squares optimization of the mixture model was performed using the lsqcurvefit method in the Optimization Toolbox of MATLAB. This video relates to the published article by J. Scott Cordova, et al., A systematic pipeline for the objective comparison of whole-brain spectroscopic MRI with histology in biopsy specimens from grade III glioma, Tomography v2(2), 2016 (
    View this media larger in a new window

  • Supplemental Media 1: Supplemental Data 1
    View this media larger in a new window

  • Supplemental Media 2: Supplemental Data 2
    View this media larger in a new window


Download the article PDF (20.66 MB)

Download the full issue PDF (76.23 MB)

Mobile-ready Flipbook

View the full issue as a flipbook (Desktop and Mobile-ready)