Schizophrenia (SZ) is a severe mental illness, with 20%-40% exhibit an inadequate or poor response to the first-line antipsychotic drugs (treatment-resistant schizophrenia, TR-SZ). However, the neural mechanisms underlying this treatment-resistance in SZ remain unclear. This study aimed to identify the co-varying multimodal pattern that distinguishing among TR-SZ, non-treatment-resistant schizophrenia (NTR-SZ) and healthy controls (HC) by unsupervised fusion, in which fractional amplitude of low-frequency fluctuation (fALFF), fraction anisotropic (FA), and gray matter volume (GMV) were used as fusion input. 63 TR-SZs, 221 NTR-SZs, and 86 healthy controls (HCs) were included in this study. The joint multimodal components (including fALFF_IC3, FA_IC12, GMV_IC9) were identified that group discriminating among TR-SZ, NTR-SZ and HC (TR-SZ < NTR-SZ < HC), constructing the central executive network (CEN) and the salience network (SAN). Moreover, these components were correlated with the cognitive scores that were validated on three independent SZ cohorts.