Probing Microstructure of Brain Tumor: A NODDI Study
Shin Tai Chong1, Hung-Wen Kao2, Chien-Yuan Eddy Lin3,4, Chiao-Chi Chen5, Ching-Po Lin1, and Chen Chang5

1Institute of Neuroscience, National Yang-Ming University, Taipei city, Taiwan, 2Department of radiology, Tri-Service General Hospital, National Defense Medical Center, Taiwan, 3GE Healthcare, Taipei, Taiwan, 4GE Healthcare MR Research China, Beijing, People's Republic of China, 5Institute of Biomedical Sciences, Academia Sinica, Taiwan


We hypothesized that the model-based diffusion model, Neurite Orientation Dispersion and Density Imaging (NODDI), could help probing microstructures of brain tumos in different types and gradings. The model perfomed better than a conventional diffusion model in differetiating meningiomas and gliomas in different grades while the two models showed significant differences of the diffusion indices between meningiomas and gliomas. In the NODDI analysis, both VFic and VFec show great potential to probe the cellularity of brain tumors.


Diffusion weighted imaging (DWI) and apparent diffusion coefficient (ADC) measurements have been widely used in the diagnosis of brain tumors and projection of the cellularity [1]. Increased cellularity results in a restricted extracellular space and thus decreased ADC values. Although the ADC values could serve as an informative indicator of the cellularity, the interpretation should be put with caution. In addition to cellularity, various factors in the heterogeneous microenvironment of the tumor might affect the ADC value, leading to inconsistent results [2]. A recently proposed model-based diffusion model, Neurite Orientation Dispersion and Density Imaging (NODDI), is promising in characterizing microstructures of the brain tissue [3]. This multi-compartment model considering intra- and extracellular as well as isotropic space could provide unique contrast across the brain tumor, which is different from that on conventional diffusion models [4]. In this study, NODDI model was used to probe microstructures of brain tumors of different types and gradings.


A total of 16 patients with brain tumors (7 WHO grade I meningiomas, 3 WHO grade II meningiomas, 3 WHO grade II/III gliomas and 3 glioblastomas) underwent 3T MRI scanning (Discovery MR 750, GE Healthcare, Milwaukee, USA) with a diffusion scheme of multiple b values in 99 directions (TR/TE = 6000/88 ms, maximal b-value = 2000s/mm2, voxel size = 2.5x2.5x2.5 mm3). All DWI images were processed with motion and distortion correction by FSL (an open source software for neuroimaging analysis, https://fsl.fmrib.ox.ac.uk/fsl/fslwiki) and fitted into NODDI model (processed with matlab NODDI toolbox, http://mig.cs.ucl.ac.uk/mig/mig/index.php/?n=Tutorial.NODDImatlab/). Three effective volume fraction (VF) of NODDI (intracellular VF, VFic, extracellular VF, VFec and isotropic VF, VFiso; where VFic + VFec + VFiso = 1) and ADC values were calculated. Region of interests (ROIs) were manually drawn around the solid part of the tumor by an experienced neuroradiologist. Since the tumor sizes were different, we randomly selected an equal number of voxels in the ROIs. One way ANOVA with post hoc analyses were employed to compare NODDI VFs and ADC values among different types and grades of brain tumors. Significance was assigned at 2-sided P value < .05.


NODDI provided a good fit in our dataset (Figure 1). In a conventional diffusion model, ADC values of meningiomas were significantly lower than those of giomas (Figure 2a). In the NODDI model, the two groups showed significant differences in VFic and VFec. In ANOVA with post hoc analyses, the Fic of WHO grade II meningiomas was marginally higher than that of WHO grade I meningiomas (Fisher's least significant differences, LSD, P = .059; Figure 2b). Besides, VFec was significantly lower in glioblastoma groups than in the glioma of lower grades (LSD, P <.001; Figure 2c). Glioblastoma showed significantly higher VFiso as compared with other groups of different tumor types and grades (Figure 2d). All the NODDI indices and mean VF were plotted in a 2-dimensional barycentric coordinate system to visualize distributions of the VFs of different groups (Figure 3).


Our results suggested that the model-based diffusion model, NODDI, could provide different microstructure information of brain tumors in comparison with conventional diffusion models. The different VFs between different grades of glioma might help accurate diagnoses. The relationship among the VF, tumor microenvironment and cellularity should be verified in histopathological study.


Both VFic and VFec show great potential to probe the cellularity of brain tumors. The NODDI analysis might help to probe different microstructures in different tumor types and gradings.


No acknowledgement found.


1. Surov A, Ginat DT, Sanverdi E, Lim CCT, Hakyemez B, Yogi A, et al. Use of Diffusion Weighted Imaging in Differentiating Between Maligant and Benign Meningiomas. A Multicenter Analysis. World Neurosurgery. 2016;88:598–602.

2. Santelli L, Ramondo G, Puppa Della A, Ermani M, Scienza R, d’Avella D, and Manara R. Diffusion-weighted imaging does not predict histological grading in meningiomas. Acta Neurochirurgica. 2010;152(8): 1315–1319.

3. Zhang H, Schneider T, Wheeler-Kingshott CA, and Alexander DC. NODDI: Practical in vivo neurite orientation dispersion and density imaging of the human brain. NeuroImage. 2012;61(4):1000–1016.

4. Wen Q, Kelley DAC, Banerjee S, Lupo JM, Chang SM, Xu D, et al. Clinically feasible NODDI characterization of glioma using multiband EPI at 7 T. Ynicl. NeuroImage. 2015;9:291–299.


Figure 1. Fits of the NODDI in white matter(anisotropic diffusion) and brain tumor(isotropic diffusion). The asterisks symbols represent the measured data in all diffusion directions with different b-values and the lines show the corresponding fits by the model. NODDI provides a good fit in whether white matter or solid tumor.

Figure 2. ADC, VFic and VFec were found significant differences between meningiomas and gliomas groups(red asterisks, P<.000). With post hoc analyses, low grade meningiomas has marginally lower VFic(LSD, P<.059) in meningiomas group(b) and Glioblastomas has significantly decrease in VFec(LSD, P<.000) in gliomas group(c). Furthermore, Glioblastomas showed significantly increased in VFiso(d).

Figure 3. Mean volume fraction of NODDI indices in each tumor groups(left). A visual comparison of NODDI indices in each groups shown with different distribution. Three NODDI indices were plotted in barycentric coordinate system(right).

Proc. Intl. Soc. Mag. Reson. Med. 25 (2017)