基于改进Fast ICA-BP算法在直肠生理信号处理中的研究

Research on the improved fast ICA-BP algorithm in rectal physiological signal processing

大便失禁是指肛门括约肌失去对粪便及气体排出的控制能力。随着医工学科的交叉发展,提出了使用人工肛门括约肌系统(AAS)来帮助患者治疗大便失禁,重建直肠感知功能,监测直肠压力和诊断直肠病变。本文改进了原括约肌执行机构,分析了直肠受力情况,进行了AAS系统的离体信号采集实验。针对AAS系统在直肠生理信号处理方面的缺失,提出了使用相空间重构(PSR)对一维直肠压力信号进行多维重建,并利用改进的快速独立分量分析(Fast ICA)算法对重构信号进行分离,提取分离后直肠信号的特征分量,采用BP神经网络进行模式识别。实验对比表明,改进的执行机构对直肠平均最大应力基本保持在12 kPa左右,保证了直肠受压区域的正常供血,实现直肠内容物的夹持。该算法迭代次数少,分离效率高,直肠功能平均诊断率达89.5%,可以实现直肠生理信号的初步处理。

Fecal incontinence refers to the ability of the anal sphincter to lose control of feces and gas discharge. With the inter-discipline development of medicine and engineering, the use of artificial anal sphincter system(AAS) has been proposed to help patients cure fecal incontinence, reconstruct rectal perception, monitor rectal pressure and diagnose rectal lesions. In this paper, the actuator of the original sphincter system is improved, the force of the rectal is analyzed, and the in vitro signal acquisition experiment of the AAS system is performed. Aiming at the lack of AAS system in rectal physiological signal processing, a multi-dimensional reconstruction of one-dimensional rectal pressure signal using phase space reconstruction(PSR) was proposed. An improved Fast Independent Component Analysis(Fast ICA) algorithm is used to separate the reconstructed signals. The characteristic components of the rectal signal after separation are extracted. And the back propagation(BP) neural network is used for pattern recognition. Experimental comparison shows that the average maximum stress of the rectum by the improved actuator is basically maintained at about 12 kPa, which ensures the normal blood supply of the rectum compression area and enables clamping of rectal contents. The algorithm has few iterations and high separation efficiency. The average diagnosis rate of rectal function can reach 89.5%, which can realize the preliminary processing of rectal physiological signals.