本文選題：調(diào)強放射治療計劃 + 相似病例檢索�。粎⒖迹�《南京航空航天大學》2017年碩士論文

【摘要】：調(diào)強放射治療(Intensity-modulated radiation therapy,IMRT)作為一種精確的放射治療技術,已經(jīng)成為目前臨床上廣泛應用的放療技術之一。計劃設計是IMRT技術的重要環(huán)節(jié),其目標是尋找靶區(qū)高劑量覆蓋與正常組織低劑量之間的最佳折中。目前IMRT計劃設計仍然是一個反復試誤且其質(zhì)量依賴于物理師經(jīng)驗的過程,由于該過程涉及的參數(shù)較多,導致計劃設計耗時較嚴重、難以保證計劃的高質(zhì)量。因此,在保證計劃質(zhì)量的同時,如何提高IMRT計劃制定效率成為目前放療研究的熱點之一。最近的研究表明,基于先驗知識放療計劃設計可有效保證IMRT計劃的高質(zhì)量、提高放療計劃的效率。本文針對基于OVH描述子的IMRT放療計劃檢索技術進行了深入的研究。首先,詳細介紹了臨床應用中IMRT計劃設計的步驟及IMRT計劃檢索技術;其次,深入研究重疊體積直方圖(Overlap Volume Histogram,OVH)描述子的幾何關系描述能力,提出基于形態(tài)學的OVH描述子計算方法,有效準確地計算OVH描述子;然后,提出基于互補OVH描述子的改進型相似病例檢索方法,該方法能夠檢索出形狀相似度更高的病例;最后,提出基于K-均值的放療知識庫聚類方法,使同種類別中的病例具有較大的相似度,通過計算待檢索病例的所屬類別進行類內(nèi)檢索,從而提高相似病例的檢索效率。為驗證所提出方法的可靠性和實用性,本文利用臨床鼻咽癌病例和前列腺癌病例,詳細分析了基于OVH描述子的IMRT計劃檢索方法的檢索性能和質(zhì)量控制性能。實驗結果表明:1)OVH描述子能夠有效地描述靶區(qū)和各危及器官間的空間位置關系,且能夠檢索到相似度較高的病例;2)利用相似病例的計劃設計參數(shù)對新病例進行再優(yōu)化,在不改變靶區(qū)劑量覆蓋率的情況下,降低了危及器官的劑量,得到的放療計劃可以與資深物理師經(jīng)反復優(yōu)化得到的放療計劃相媲美,實現(xiàn)了對放療計劃質(zhì)量的控制;3)用相似病例指導計劃設計,避免了計劃優(yōu)化的反復試誤過程,為提高計劃設計效率奠定基礎。
[Abstract]:Intensity-modulated radiation therapy (IMRT), as a precise radiotherapy technique, has become one of the widely used radiotherapy techniques in clinic.Planning design is an important part of IMRT technology, whose goal is to find the best compromise between high dose coverage of target area and low dose of normal tissue.At present, the design of IMRT plan is still a process of repeated trial and error and its quality depends on the experience of the physicist. Due to the large number of parameters involved in the process, the planning design is time-consuming and difficult to guarantee the high quality of the plan.Therefore, how to improve the efficiency of IMRT planning becomes one of the hotspots in radiotherapy research.Recent studies have shown that the design of radiotherapy plan based on prior knowledge can effectively guarantee the high quality of IMRT plan and improve the efficiency of radiotherapy plan.In this paper, IMRT radiotherapy plan retrieval technology based on OVH descriptor is studied.Firstly, the steps of IMRT plan design and IMRT plan retrieval technology in clinical application are introduced in detail. Secondly, the geometric relation description ability of overlapped Volume histogram descriptor is studied in depth, and the method of calculating OVH descriptor based on morphology is proposed.Then, an improved similar case retrieval method based on complementary OVH descriptor is proposed, which can retrieve cases with higher shape similarity.In this paper, a method based on K-means is proposed to cluster the knowledge base of radiotherapy, which makes the cases in the same category have greater similarity. The retrieval efficiency of similar cases can be improved by calculating the category of the case to be retrieved.In order to verify the reliability and practicability of the proposed method, the retrieval performance and quality control performance of IMRT plan retrieval method based on OVH descriptor were analyzed in detail by using clinical cases of nasopharyngeal carcinoma and prostate cancer.The experimental results show that the 1: 1 OVH descriptor can effectively describe the spatial position relationship between the target area and the organs at risk, and can retrieve the case with high similarity.) the new case can be reoptimized by using the planning design parameters of the similar case.Without changing the dose coverage of the target area, the dose at risk to the organ was reduced, and the resulting radiotherapy plan was comparable to that obtained by repeated optimization by a senior physicist.It realizes the quality control of radiotherapy plan and uses similar cases to guide the plan design, avoids the repeated trial and error process of the plan optimization, and lays the foundation for improving the efficiency of the plan design.
【學位授予單位】：南京航空航天大學
【學位級別】：碩士
【學位授予年份】：2017
【分類號】：R730.55;TP391.41

【參考文獻】

相關期刊論文 前10條

1 李志偉;;機器學習發(fā)展及幾種學習方式[J];產(chǎn)業(yè)與科技論壇;2016年10期

2 陳萬青;鄭榮壽;張思維;曾紅梅;左婷婷;賈漫漫;夏昌發(fā);鄒小農(nóng);赫捷;;2012年中國惡性腫瘤發(fā)病和死亡分析[J];中國腫瘤;2016年01期

3 陳軼;;機器學習技術在醫(yī)療數(shù)據(jù)挖掘中的應用[J];電子測試;2015年03期

4 陳萬青;鄭榮壽;曾紅梅;鄒小農(nóng);張思維;赫捷;;2011年中國惡性腫瘤發(fā)病和死亡分析[J];中國腫瘤;2015年01期

5 石國勇;楊曉;李巖;;機器學習算法在綠色智能化汽車發(fā)展及研究中的應用展望[J];科技創(chuàng)新導報;2014年04期

6 趙平;;中國癌癥流行態(tài)勢與對策[J];醫(yī)學研究雜志;2013年10期

7 秦亮;張文廣;史賢俊;肖支才;;基于免疫克隆算法的多目標聚類方法[J];信息與控制;2013年01期

8 陳萬青;張思維;鄭榮壽;曾紅梅;鄒小農(nóng);趙平;吳良有;李光琳;赫捷;;中國2009年惡性腫瘤發(fā)病和死亡分析[J];中國腫瘤;2013年01期

9 謝娟英;郭文娟;謝維信;高新波;;基于樣本空間分布密度的初始聚類中心優(yōu)化K-均值算法[J];計算機應用研究;2012年03期

10 鄭榮壽;張思維;吳良有;李光琳;趙平;赫捷;陳萬青;;中國腫瘤登記地區(qū)2008年惡性腫瘤發(fā)病和死亡分析[J];中國腫瘤;2012年01期

相關碩士學位論文 前1條

1 李宇泊;K均值算法初始聚類中心選取相關問題研究[D];蘭州交通大學;2012年

，

本文編號：1730943