A Study Of Knowledge-based Radiation Optimization For Automatic Radiation Planning And Clinical Application

In modern society,malignant tumors were the major diseases which threaten humem's healthy.Radiotherapy,chemotherapy and surgery were the main treatments for cancer.Overall,the treatment goal of radiotherapy technology is to maximize radiation doses in the malignant tumor area(tumor target area),thereby killing tumor cells.In the same time,the surrounding normal tissues were protected from harming and toxicity effect of these organ were reduced.In this case,the effectiveness of radiation therapy would be improved.With the continuous accumulation of radiation clinical cases,there are a large number of cancer patients' image information and dose planning information in current radiotherapy systems.This information becomes a priori knowledge.The feature extraction and quantitative analysis of these prior knowledge(knowledge-based-system)can realize the intelligence of the radiotherapy planning system,so as to effectively improve the accuracy of radiotherapy and predict the efficacy of radiotherapy.Main contents in this paper are as follows: 1)Knowledge-based optimized algorithms of radiation therapy: resulting from the different principle of DVO and DMPO,it might not sufficient and effectiveness that only one optimized algorithm were implemented of online ART.Although several comparisons study between these two algorithm to improve treatment plan quality to the best of our knowledge,there are no studies comparing the optimized efficiency of two algorithms on the online ART or classifying target deformation in inter-fraction.In addition,we aimed to investigate whether the biological optimization could be an alternative method to the conventional physical optimization for cervical carcinoma,by comparing plan qualities be-tween the dose volume based optimization(DVO,only physical constraints were used in the optimization)and biological model based optimization(BMO,biological constraints were used in the optimization with several assistant physical constraints).2)Knowledge-based method guided hydrogel injection process: firstly,we have demonstrated the feasibility of an endoscopic ultrasound-guided injectable hydrogel separation technique using a cadaveric model to increase the space between the head of the pancreas and duodenum.Using modeling studies,we identified the minimum distance of this separation for optimal sparing of the duodenum,setting the foundation for future clinical trials using this technique to enable dose escalation with either stereotactic or intensity-modulated radiation therapy for patients with unresectable pancreatic cancer.Secondly,we established an OVH prediction model to guide the selection of pancreatic cancer patients who may benefit from hydrogel spacer injection before radiation for duodenum sparing.We evaluated the model's accuracy by verifying predictions of the required hydrogel spacer thickness to achieve clinical dose constraints.Our upcoming clinical trials can utilize this technique to ensure sufficient separation between duodenum and head of the pancreas for consideration of dose-escalated radiation therapy.In the end,because biologically effective dose(BED)greater 70 Gy has been reported to have a superior overall survival in pancreatic cancer treatment,we further aimed to overcome the limited prescription dose caused by the dose constraint of adjacent normal tissues,specifically duodenum when applying dose escalation with stereotactic body radiotherapy(SBRT).