P14.09.A DIFFUSION MICROSTRUCTURE IMAGING AS A DIAGNOSTIC CORRELATE OF HISTOMORPHOLOGY: ASSESSMENT ACROSS THE MOST COMMON INTRACRANIAL TUMOR ENTITIES

Abstract BACKGROUND Distinct histomorphologic features differentiate the four most common intracranial neoplasms: meningioma, metastasis, glioblastoma, and primary CNS lymphoma. Utilizing advanced diffusion imaging, we explore how non-invasive assessment of microstructural tumor characteristics can provide insights into intratumoral morphology and may guide further diagnostic strategies. MATERIAL AND METHODS 101 newly diagnosed intracranial tumors (35 metastases, 34 glioblastomas, 21 meningiomas, 11 primary CNS lymphomas) underwent advanced diffusion imaging including Diffusion Tensor Imaging (DTI), Neurite Orientation and Dispersion Imaging (NODDI), and Diffusion microstructure Imaging (DMI) at 3 Tesla. In addition to DTI-derived estimates about the directionality and extent of diffusion, the biophysically motivated diffusion NODDI and DMI allowed subdivision of voxels into an intra-axonal (NODDI ICVF, DMI V-intra), extra-axonal cellular (DMI V-extra) and a free water (NODDI ISO-VF, DMI V-CSF) fraction using a multi-compartment model. We evaluated segmented contrast-enhancing tumor areas in relation to confirmed definitive diagnoses and histomorphology based on correlated biopsy material. Ultrahigh resolution microstructure analysis of selected pre-fixated tumor tissue samples at 9.4 Tesla is used to test association of distinct diffusion parameters with histomorphological characteristics of tissue compartments, heterogeneity, directionality, axonal preservation and cell type. RESULTS Pairwise comparison of tumor entities identified distinct diffusion metrics encompassing the different diffusion techniques as ideal discriminators for each two tumor entities. Differentiation of glioblastoma and meningioma was best achieved by the intracellular compartment (DMI V-intra, p = 0.005) and NODDI-ICVF (p = 0.0008). Likewise, glioblastoma and metastasis were discriminated by the extracellular compartment (DMI V-extra, p = 0.002) and DTI-OD (p = 0.00008[DTD1] ), glioblastoma and primary CNS lymphoma by the CSF compartment (DMI V-CSF, p = 0.002) and DTI-aD (p = 0.00005), meningioma and metastasis by the CSF compartment (DMI V-CSF, p = 0.0002) and NODDI-ISOVF (p = 0.0007), and metastasis and primary CNS lymphoma by the CSF compartment (DMI V-CSF, 0.00004) and NODDI-ISOVF (p < 0.00001). Spatial distribution of the DMI discriminators within tissue blocks were correlated with histomorphological features at ultrahigh resolution. CONCLUSION Multiparametric diffusion imagi