Brain Structure Analysis Based On Cancer Radiotherapy

Cancer has become a serious disease all over the world.Its pathogenesis and the characteristics of metastasis easily make cancer treatment difficult and inefficient.As an important means of cancer treatment and prognosis,radiation therapy plays an important role in adjuvant treatment and prevention of cancer recurrence and metastasis.However,while radiotherapy effectively controlled cancer cells,it also caused damage to peripheral cells.Long-term follow-up has found that cognitive and mental disorders are common in patients.In addition,the best radiotherapy scheme and the choice of the timing of radiotherapy can effectively control the cancer cells.There was a corresponding radiotherapy program in different stages,and the rational choice of the best radiotherapy time and scheme can effectively improve the survival rate of the patients.Therefore,in this paper,based on the structural magnetic resonance imaging(sMRI)data,the cerebral gray matter morphology index as the starting point,from four aspects that were the effect of radiotherapy,the radiation-induced injury,the influencing factors of radiotherapy program-cancer staging,the timing of radiotherapy,in-depth analysis of the relationship between these aspects of radiotherapy and the brain gray matter,help to further fully understand all aspects of radiotherapy.The main work was as follows:1.Based on a typical head and neck tumor of nasopharyngeal carcinoma(NPC),the relationship between radiotherapy effect,radiation injury and brain structure was studied.First,the voxel based morphological analysis(VBM)was used to study the changes in the volume of gray matter in the before and after radiotherapy.Secondly,the brain structure network was constructed,and then the topological properties and node characteristics of the brain structure network before and after radiotherapy were analyzed by graph theory.Finally,based on a simple machine learning framework,a model for automatic classification of before and after radiotherapy was built.Results: after radiotherapy,the gray matter volume of NPC patients significantly changed;The both networks before and after radiotherapy all showed the small world property.Compared with before radiotherapy,the brain structure network after radiotherapy had higher global and local efficiency,shorter path length and larger clustering coefficient.The correct rate of classification before and after radiotherapy was 82.5%.2.Based on the different stages of T staging of NPC,the stages of the influencing factors of the radiotherapy plan were classified and studied.First,the indexes correlated with gray matter in each stage were analyzed quantitatively.Secondly,different T stages were classified using a simple machine learning framework.The results show that the indexes correlated with gray matter significantly changed.The classification results between all pairs of four stages showed that the classification accuracy between T2,T3 and T4 were low,and the classification effect between T1 and T2,T3 and T4 were better.The correct rate of classification between early stage(T1)and late stage(T2-T4)was 80.77%.3.With the common and high incidence lung cancer patients with brain metastases as the research object,the time of brain metastasis was predicted.First,the gray matter images before and after brain metastases were statistically analyzed.Secondly,the multiple regression model was used to construct the prediction model of the time of brain metastasis form lung cancer.Results: after brain metastasis,gray matter volume significantly changed;Prediction results showed that there was a good correlation between the estimated interval time and the original interval time(r=0.77,p=3.03e-9).