Prognostic Analysis Of Tumor With Multimodal Data Fusion

As an important branch of precision medicine,prognosis prediction of malignant tumors can assist individual decision-making of tumor diagnosis and treatment.How to effectively use the multi-modal data obtained from the current diagnosis and treatment system to guide the prognosis assessment of cancer is one of the important contents of current cancer research.This thesis focused on the prognostic prediction of three common types of cancer(i.e.,clear cell renal cell carcinoma,glioma of brain,cervical cancer),analyzed the key problems faced by them,and used the obtained multimodal data to establish precise multimodal prognostic prediction models.The main research contents of this thesis are as follows:(1)Based on the analysis of transcriptome data,computer tomography(CT)image data,pathological image data and follow-up data of 209 patients with renal cancer in public database,the deep features of genome,image group and pathological group are extracted.We explored the correlation between CT/pathological image features and gene pathway modules,and established a multimodal prognosis prediction model for renal cancer.(2)By analyzing the Magnetic resonance imaging(MRI)image data of 214 cases of glioma in the public database,the radiomics and deep features of tumor and surrounding edema necrotic area were extracted,and the performance of the models constructed based on different image features was analyzed and compared.A multimodal prognostic model for glioma was established.(3)By analyzing the CT/MRI of 104 cervical cancer patients undergoing radical radiation therapy in Nanfang hospital,the characteristics of cervical tumor regions were extracted,different prediction models were constructed,and the prediction performance and units of different models were compared.Finally,a multimodal prognosis prediction model based on Transformer network for cervical cancer was constructed.The results showed that the prognostic prediction model established based on multimodal data could stably and effectively stratify the prognostic risk of tumor patients.It has great potential for clinical application.