Virtual Reality Assistant System For Liver Cancer Operation And Study On The Prognosis Of Hepatocellular Carcinoma

Primary liver cancer is a common malignant tumor in China.Its morbidity occupies the fourth position among malignant tumors,and its mortality ranks second.Hepatocellular carcinoma accounts for 85% to 90% of primary liver cancer.Currently,the multidisciplinary approach based on surgery is the primary therapy for liver cancer.During the surgery,it should be guaranteed that the remaining functional liver volume is maximized and prevents the accidental injury of the liver duct,under the condition that the tumor tissue is completely removed.Thus,the structure of a patient’s liver should be accurately understood.The fiveyear survival rate of patients with hepatocellular carcinoma is only 10% to 20%.This is largely because hepatocellular carcinoma has the characteristics of easy metastasis and recurrence.Postoperative recurrence severely limits the long-term survival of patients with hepatocellular carcinoma.Therefore,performing prognostic analysis of hepatocellular carcinoma and exploring clinical indicators associated with postoperative recurrence can help improve the survivability patients with prognosis.Aiming at the construction of a virtual reality assisted system for liver cancer surgery,the prognostic survival analysis of hepatocellular carcinoma,and the prediction of metastasis and recurrence after surgery,this thesis carries out the relevant research by combining several emerging techniques such as virtual reality and machine learning.It is divided into three parts.In the first part of the work,we apply deep learning and virtual reality technology to the preoperative assistance of liver cancer resection,constructing a preoperative assistance system of liver cancer resection.The virtual reality software is developed to realize the interaction between doctors and patient organs.Thus,doctors can immersively check the degree of liver lesions,the adhesion of tumor to surrounding blood vessels,bile ducts and other tissues.The second part of the work is the prognosis survival analysis of hepatocellular carcinoma based on the Cox model.For this work,the utilized data is from the TCGA database.A prognostic survival model is constructed by selecting the patient clinical indicators as the risk factors,which analyzes the effects of each clinical indicator on the patient’s prognostic survival.In the light of the model,the nomogram is then plotted,which is used as a prognostic survival analysis tool.The third part of this thesis is the analysis of postoperative recurrence prediction of primary hepatocellular carcinoma based on the XGBoost model.Similarly,the data used in this work is from the TCGA database.Using the XGBoost model for feature selection and recurrence prediction,the clinical indicators closely related to the postoperative recurrence are filtered out;the performance of the prediction model on the test set is also tested.The obtained achievements are as follows.The liver and tumor are segmented from CT images of liver cancer patients.In combination with the patient’s three-dimensional organ model,a virtual reality-based liver cancer surgery assistant software has been developed,which realized functions such as organ segment inspection,virtual endoscope,and internal inspection of organs.It provides a promising solution for clinical application scenarios such as preoperative three-dimensional inspection and remote consultation.For prognostic survival analysis of hepatocellular carcinoma,a prognostic survival model has been constructed based on Cox multivariate regression analysis and a nomogram has been obtained.The consistency index of the model on the test set is 0.682(95% CI: 0.558-0.746).The evaluation of the calibration curve shows that the model well fits the patient’s actual survival rate;the model has a certain ability to predict the survival of the patient’s prognosis.Moreover,the risk factors that have a great impact on the model have been analyzed in the discussion.For the research on the prediction of postoperative recurrence of primary hepatocellular carcinoma,the accuracy rate of the overall prediction model is 0.75,the sensitivity and specificity are 0.63 and 0.83 respectively,and the value of AUC is 0.78.In addition,several clinical indicators closely related to the postoperative recurrence have been screened out and discussed in detail.In summary,this thesis investigates the surgical assistance and the prognosis of hepatocellular carcinoma in liver cancer treatment,which provides a guidance for the surgical treatment and the formulation of postoperative recovery plans for liver cancer in clinical.