Abstract: Objective To examine the correlation of apparent diffusion coefficient(ADC), diffusion weighted imaging(DWI), and T1 contrast enhanced(T1-CE)with Ki-67 labeling index(LI)in primary central nervous system lymphomas(PCNSL), and to assess the diagnostic performance of MRI radiomics-based models in differentiating the high-proliferation and low-proliferation groups of PCNSL. Methods MRI images and clinical information of 83 PCNSL patients were included, and their correlation with Ki-67 LI was examined using Spearman correlation analysis. The imaging histological features of three sequences were extracted separately and seven different imaging histological models were constructed. The receiver operating characteristic(ROC)curve was used to evaluate the predictive performance of all models. Delong test was utilised to compare the differences of models. Results Relative mean ADC(rADCmean)(ρ=-0.354, P=0.019), relative mean DWI(rDWImean)(b=1,000)(ρ=0.273, P=0.013)and relative mean T1-CE(rT1-CEmean)(ρ=0.385, P=0.001)were significantly correlated with Ki-67. The best prediction model is ADC+DWI+T1-CE(AUC=0.869). Conclusion rDWImean, rADCmean and rT1-CEmean are correlated with Ki-67 LI. The radiomics model based on MRI sequences combination is promising to distinguish low proliferation PCNSL from high proliferation PCNSL.

Key words: Primary central nervous system lymphomas, Radiomics, Multi-parameter, Ki-67 labeling index, Magnetic resonance imaging