Metastatic brain tumours and brain glioblastomas are malignant by nature and account an absolute majority of all brain tumours in adults. The metastasis and glioblastomas present around 50% and 20% of all intracranial tumours, respectively¹. A primary glioblastoma constitutes 90% of all glioblastomas, whereas less than 10% of the glioblastomas are secondary. The secondary glioblastomas originate from WHO grade II and grade III gliomas². The primary glioblastoma is more aggressive and has much worse prognosis compared to secondary one1. Conventional MRI allows us to differentiate the glioblastoma types. The primary glioblastomas are characterised by a contrast enhancement of whole solid tumour, tumour necrosis, and peritumoral brain edema. The secondary glioblastomas have large nonenhancing solid tumour regions regions without a contrast enhancement and rarely undergo necrosis (see Fig.1) Often it is difficult to differentiate the primary glioblastomas from the brain metastasis using the conventional MRI since both tumour types demonstrate similar radiological features. In our work we assessed the diagnostic efficacy of the diffusion kurtosis imaging in a differentiation of the brain metastasis and primary glioblastomas.To assess the efficacy of diffusion kurtosis imaging in a structural differentiation of the brain metastasis and primary glioblastomas.20 patients with the brain metastasis (5 with a breast cancer, 6 with a lung cancer, 5 with a colorectal cancer, and 4 with melanomas) and 35 patients with primary glioblastomas underwent 3T MR imaging and diffusion weighted measurements (b­values = 0, 1000 and 2500 s/mm², 60 gradient directions, resolution = 3mm³). Diffusion scalar metrics such as mean kurtosis (MK), axial kurtosis (AK), radial kurtosis (RK), kurtosis anisotropy (KA), mean diffusivity (MD), axial diffusivity (AD), radial diffusivity (RD), fractional anisotropy (FA), relative anisotropy (RA), intracellular water fraction (ICWF), axial extracellular diffusivity (AECD), radial extracellular diffusivity (RECD), tortuosity (Tort) were estimated and later compared with solid regions of metastasis and primary glioblastomas using a criterion of significant level p<0.05 and a Mann­Whitney test. A tumour necrosis and peritumoral brain edema were excluded from the regions of interest. All primary glioblastomas were newly diagnosed, without having previous radiation or chemotherapy. All brain metastasis also were newly diagnosed, without having brain previous radiation. The diagnosis in all cases were confirmed by a tumour removal or stereotaxic biopsy performance in 1-2 weeks after MRI screening.Values of MK, Tort, FA and RA did not show the significant difference between the brain metastasis and primary glioblastomas. The AK, RK, KA and ICWF were significantly higher (p<0,05) in the primary glioblastomas comparing to metastasis. The values of MD, AD, RD, AECD and RECD were significantly higher in metastasis comparing to primary glioblastomas (see Fig. 2).An application of non-Gaussian diffusion models to a division of different tumour types in the human brain is very challenging problem. On the one hand, the microstructure organisation of tumour cells has a significant difference in a water diffusion dynamics due to a cell compartmentalization comparing to a normal white matter tissue with the ordered axon bundles. On the other hand, we expect to see the difference in the tumour cell organisation based on the tumour origin: primary or metastatic cancer what can be seen by application of contrast enhancement technique. In a statistical comparison of primary and metastasis tumours we found the higher values of AK, RK, KA, ICWF and lower values of mean diffusivity MD, AD, RD, AECD and RECD in the primary glioblastomas compared to the metastasis. These findings reflect a fact that a water diffusion in the glioblastomas is more restricted. We hypothesized that this effect might be connected to the higher cellularity of the glioblastomas in contrast to the metastatic cancer tissue. In turn, this effect can be detected and visualised by DKI metrics.Diffusion kurtosis imaging and two compartment model demonstrated a high potential in the structural differentiation of the brain metastatic tumours and primary glioblastomas. However, more detailed diffusion model³ accounting cellular nature of the tumour should be taken in account in the future.