【作者】 姜恩宇；

【导师】 朱晓锦； 昝鹏；

【作者基本信息】 上海大学 ， 控制理论与控制工程， 2013， 博士

【摘要】 大便失禁指肛门括约肌失去对粪便及气体排出的控制能力，是临床上常见的病症。它容易造成多种并发症，尤其在老年患者、危重患者及瘫痪卧床患者中发病率居高不下，不仅给患者带来极大痛苦，而且也给护理工作带来诸多困难。随着人口老龄化趋势的发展，大便失禁已成为医疗、护理亟需解决的问题，相关治疗手段与科学技术的发展对提高患者生命质量和推动社会文明进步具有重要的现实意义。治疗大便失禁的传统方法很多，但是都存在一定的缺陷，甚至可能产生严重的并发症。随着人们对健康和生命质量的要求不断提高，传统的治疗方法已经不能满足患者的需求，人工肛门括约肌为大便失禁的治疗开辟了一条新途径；国内外学者已经针对人工肛门括约肌的仿生结构、控制方式、能量供给等方面开展了大量研究工作。但是，当前人工肛门括约肌领域涉及的直肠感知功能重建的研究尚无显著进展，因此围绕目前人工肛门括约肌研究无法重建患者直肠感知功能这一关键问题，进行相关技术方法的深入探索和研究，并充分发挥现代信息处理技术、材料科学技术、生物医学技术等多学科交叉优势，研究一种完全适合植入人体的人工肛门括约肌势在必行。本文基于国家自然科学基金科研项目的研究背景，以人工肛门括约肌为研究对象，着重进行直肠感知功能重建的理论方法及相关技术的研究，同时针对特征向量选取和模式识别算法优化的相关方法进行积极探索，并在此基础上构建人工肛门括约肌直肠感知功能重建的实验平台，以验证相关方法技术的可行性和有效性；全文工作可以概括为人工肛门括约肌系统的开发和实现、直肠感知功能重建理论与方法、模拟肠道实验平台构建与验证分析三大部分。所做主要研究工作与贡献如下：（1）介绍人体排便的基本生理常识和人体肛门失禁的概况；阐述了肛门失禁治疗方法和人工肛门括约肌方面的国内外研究现状；分析了目前临床医学上针对肛门失禁的治疗方法及其不足；根据人体生物肛门和结直肠的结构特点，提出基于生物反馈控制的人工肛门括约肌系统构思。（2）人工肛门括约肌系统的仿生学实现。本论文在研究肛门括约肌肌群生理学的基础上，探索运用工程技术重建肛门括约肌理论、方法和排便控制机理，模拟肛门括约肌各肌群的基本功能；提出系统技术方案，分析系统三大技术模块：基于生物反馈控制的人工肛门执行机构、无线通信模块和经皮供能模块；按照模块化和低功耗的要求研究人工肛门括约肌系统的实现结构。（3）直肠压力收缩信号的特征提取算法研究。本文在分析现有特征提取方法及存在的问题基础上，针对直肠收缩信号的非线性、非平稳性，提出了基于小波包分解子空间时域特征及子空间频域特征的的直肠压力特征提取方法。为从原始特征集中找到最有利于分类的最优特征集，以提高分类器的精度和分类准确率，提出了基于Fisher距离的最优小波包基直肠压力特征选择方法，以及基于DB（Davies-Bouldin，DB）指标优化的最优小波包基直肠压力特征选择方法，并基于概率神经网络（Probilistic Neural Network，PNN）模型验证了提出方法的有效性。（4）直肠压力收缩信号的分类算法研究。本文将基于PNN的分类、基于支持向量机的分类应用于直肠收缩信号的分类预测；同时提出了基于粒子群优化算法优化支持向量机参数的直肠收缩信号特征向量分类方法，并在此基础上对采集到的直肠压力特征向量进行了分类预测，对比分析了三种分类预测方法的分类准确率，验证了所提分类方法的有效性和优越性。（5）构建基于生物反馈式人工肛门括约肌实验模型与实验平台，在上述理论与方法的指导下，通过模拟人体肠道环境，获取大量实验数据，验证人工肛门括约肌的基本功能、直肠感知功能重建模型的有效性，根据评估结果改进和完善人工肛门括约肌系统，为大便失禁患者人工器官的最终临床应用提供理论方法依据与技术基础支持。更多还原

【Abstract】 Anal incontinence is a common disease in clinic, which means natural analsphincter loses the control ability for defecation of fecal and gases. It is likely tocause a variety of complications, and the incidence is higher in elderly patients,critically ill patients, and paraplegic patients. It doesn’t only bring great suffering tothe patient, but also increases difficulties to nursing work. Under the tendency ofaging population, fecal incontinence has been addressed as a matter of urgency inmedical and nursing field. The development of related therapeutic means andtechnology has a vital practical significance to improve patients’ life quality and topromote social civilization.There are several traditional therapy treatments for anal incontinence. However,these treatments have some drawbacks and might even lead to severe complications.As the development of modern technology, people have asked for more requirementson health and quality of the life, so traditional treatments cannot meet patients’requirement. Artificial anal sphincter has opened up a new way for treatment of analincontinence. Lots of related research work aiming at the bionic mechanism, controlmanner and power supply of artificial anal sphincter have been studied by domesticand foreign scholars, while the current research work involving rectal sensationfunction rebuilding has no significant progress. So it is imperative to solve theproblem of how to rebuild rectal sensation function, and use interdisciplinary field ofmodern information processing technology, material technology and biomedicine, todesign an artificial anal sphincter system completely suitable to be implanted inhuman body.Based on the research background of National Natural Science Fund of China,the dissertation takes artificial anal sphincter as research object and focuses on therelated theory and method research of rectal sensation rebuilding. Related approachaiming at feature vector selection and pattern recognition optimizing is explored, andan integrated experimental platform of rectal sensation function rebuilding isestablished to verify its feasibility and effectiveness. The dissertation can besummarized to three main parts, which are implementation of artificial anal sphinctersystem, theory approach of rectal sensation function rebuilding and the third part includes the establish of intestinal simulations experimental platform and the theoryverify. The research work is mainly reflected as follows:(1) The dissertation introduces the basis physiological sense and the generalsituation of anal incontinence, then states the general treatment for anal incontinenceand related domestic and foreign research present situation. In view of existingproblem of traditional therapy treatment, the rectal sensation function model is built.According to the structure feature of human nature anal and colorectum, thedissertation proposes the conception of an artificial anal sphincter system based onbiological feedback control.(2) The bionics realization of artificial anal sphincter system is discussed. Usingengineering technology, the dissertation studies the nature anal sphincter musclegroup in physiology based on anal sphincter rebuilding theory, approach anddefecation control mechanism to simulate the basis function of anal sphincter musclegroup. The artificial anal sphincter system that consists of executive mechanism,wireless communication module and transcutaneous energy transmission module isproposed. The research and implementation structure of the artificial anal sphinctersystem are based on modeling and low-power consumption requirement.(3) Feature extraction algorithm of rectal pressure contraction signal is studied.The dissertation firstly studies the problem of existing feature extraction method. Inview of the non-stationary and nonlinear characteristics of rectal pressure signal,feature extraction method based on wavelet packet decomposition subspace timedomain feature and subspace frequency domain feature is proposed. In order to findthe optimal feature set from the original feature set for a higher classify accuracy andprecise, the dissertation proposes rectal pressure signal optimal wavelet packet basisfeature extraction method based on Fisher distance and optimal wavelet packet basisextraction method based on DB (Davis-Bouldin, DB) index. And the effectiveness ofthe two feature extraction methods has been verified based on PNN (ProbabilisticNeural Network, PNN) model.(4) The classify algorithm of rectal pressure contraction signal is studied. Bothof the classification based on PNN and the classification based on SVM are appliedfor the classification forecasting of rectal pressure contraction signal. Meanwhile,PSO (Particle Swarm Optimizer, PSO) algorithm is used to optimize SVMparameters, and classification forecasting is applied for the collected rectal pressure signal feature vectors. At last, contrast analysis of prediction accuracy using the threedifferent methods is studied to verify the proposed classification methods’effectiveness and superiority.(5) An experimental platform prototype of the artificial anal sphincter systembased on biological feedback is developed and human nature intestinal simulationexperiment is carried out. With the guidance of the above theory and methods, a lotof experimental data is obtained through the human nature intestinal simulationenvironment. Experimental results verify the basic function of artificial analsphincter and effectiveness of the rectal sensation function rebuilding model. Basedon the assessment result, it is possible to improve the artificial anal sphincter systemand provide theoretical basis and technical support for AI patients in clinic treatment.更多还原