Clinically distinguishing the multiple sclerosis (MS) and neuromyelitis optica spectrum disorder (NMOSD) is critical, since the prognosis and treatment of these disorders differ. We extracted nine radiomics features from 485 radiomics features combining with clinical measurements to build the model for differentiating MS and NMOSD. The area under receiver operating characteristic curve (AUC) of the model was 0.8808 and 0.7115 in the primary and validation cohort. The model demonstrated good calibration. The current study revealed the different radiomics features between MS and NMOSD, and developed and validated an individual model to differentiate the two diseases.