High-grade serous ovarian carcinoma (HGSOC) is the most common subtype of ovarian carcinoma. Platinum-based drugs are the first-line chemotherapeutic agents for HGSOC, but platinum resistance and prognosis are difficult to predict. Domestic and foreign studies have found that multi-functional MRI could reflect the macroscopic heterogeneity of tumors from different perspectives; habitat imaging contributed to reflecting the spatial heterogeneity of tumors; the attention mechanism could integrate features of different scales; and multi-omics were capable to improve predictive performance. Previously, our team has demonstrated the importance of MRI and its functional imaging in the diagnosis and histological evaluation of gynecological tumors. And conventional MRI habitat imaging, multi-instance learning based on whole slide image (WSI), and multi-omics model could effectively predict platinum resistance in HGSOC patients. Therefore, this study aims to perform habitat imaging on multi-functional MRI such as multi b-value DWI and DCE-MRI and to train WSI-based multi-instance learning to construct pathological signature. Then, combined with clinical indicators, radiomics based on MRI habitat, and pathomics, a multi-omics fusion model trained by attention mechanism was constructed. Finally, to explore the value of MRI habitat, WSI, and multi-omics in predicting platinum resistance and prognosis in HGSOC patients. This study combines macroscopic functional imaging and microscopic pathological information to construct a multi-omics complementary model, which can predict platinum resistance and prognosis of HGSOC patients more comprehensively and accurately, and further guide the formulation of clinical treatment.