Image based quantification methods for optimization of differential diagnosis of tumors in brain MRI
Loading...
Date
Authors
Βαμβακάς, Αλέξανδρος Χρυσοβαλάντης
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
Preoperative differentiation of brain tumors is critical for adapting suitable treatment strategies, as well as evaluating tumor response to therapy. However, differentiation solely relying on conventional Magnetic Resonance Imaging (MRI) remains a challenge, due to ambiguous tumor imaging appearance.
Glioblastoma Multiforme (GBM) and Solitary Metastasis (MET) are common brain tumors, which represent a characteristic example of such a diagnostic problem. Although histopathological analysis of biopsy samples is the gold standard for establishing diagnosis, it is not always feasible, due to its invasive character, while it samples only a limited portion of the lesion tissue. This has led to the exploitation of a variety of advanced MRI techniques, which provide structural and functional information of tumors, such as Diffusion Tensor Imaging (DTI). In addition to advanced MRI techniques, image analysis methods have been recently introduced regarding exploitation of quantitative tumor imaging parameters, in terms of tumor morphology and texture.
In this study, 3D tumor models are generated by a DTI clustering segmentation technique, providing up to 16 brain tissue diffusivities, complemented by post-contrast T1 imaging, resulting in the identification of tumor core. Subsequently, 3D surface is refined by a Morphological Morphing interpolation technique, for reducing noise induced by anisotropic voxel size. The 3D models are analyzed in terms of their surface and internal signal variations characteristics towards identification of discriminant features between GBMs and METs, utilizing a case sample of 10 GBMs and 10 METs.
Morphology analysis of tumor core surface is assessed by 5 local curvature features. Texture analysis considers 11 first and 16 second order 3D textural features. From the 16 second order features, 11 are based on Gray Level Co-Occurrence Matrices (GLCM) and 5 on Gray Level Run Length Matrices (GLRLM), calculated from DTI isotropic and anisotropic parametric maps corresponding to 3D active region segments of tumor core, resulting from the clustering technique, considering 3 different quantization levels (QL) and 1-4 pixels displacements (D).
Case sample distributions of morphology and texture features were analyzed using the Mann-Whitney U test, with a cut-off value of 0.05 to identify discriminant features. Results highlight the value of all 5 local curvature descriptors to capture differentiation between GBMs and METs. Also 2 out of 11 first order features, regarding isotropy maps were useful, while 7 out of the 11 GLCM features were capable of discriminating heterogeneity of anisotropic diffusion properties of GBMs and METs, at QL=6 and D=2. Finally, all 5 GLRLM features extracted from diffusion isotropy maps seem to discriminate structural properties of GBMs and METs, at QL=5.
Results demonstrate the potential of surface morphology and texture analysis of imaging appearance of 3D tumor models in pre-treatment brain MRI tumor differential diagnosis.
Description
Keywords
Brain tumors, Diffusion tensor imaging, Magnetic resonance imaging, Texture analysis, Image analysis, Glioblastoma multiforme, Solitary metastasis, Local curvature analysis, Differential diagnosis