 Advances in Technology Innovation , vol. 2, no. 2, 2017, pp. 46 - 50 46 Assessment of Arteriovenous Shunt Pathway Function and Hypervolemia for Hemodialysis Patients by Using Integrated Rapid Screening System Wei-Ling Chen 1 , Chia-Hung Lin 1 , Chung-Dann Kan 2,* 1 Department of Engineering and Ma intenance, Kaohsiung Veterans Genera l Hospital, Kaohsiung, Taiwan. 2 Division of Cardiovascular Surgery, Department of Surgery, National Cheng Kung University Hospital, College of Medicine, National Cheng Kung University, Tainan, Taiwan. Received 06 January 2016; received in revised form 25 March 2016; accept ed 01 April 2016 Abstract Currently, the hemodialysis patients received body weight measurement by themselves, vita l sign checking by nursing staffs before dia lysis. Whenever, the arteriovenous routes with problems doubted, the patients needed to be referred to sur- geon for vascular echography checking and then to be corrected. How to integrate these three tasks in one time is a very important issue. The project proposes to combine our previous study of au- dio-phono angiographic technology in detecting vascular stenosis with rapid screening system to evaluate dialysis patients’ arteriovenous routes function and their status of e xcess body fluids: inspecting and integrating the blood pressure, body weight, and fistula function work into a rapid screening system, and using the quantiza- tion of fistula phono angiography pitch to achieve assessing arteriovenous routes. Future hoping is developed a complete integrated intelli- gence system by combining the arteriovenous fis- tula signal processing with feature extraction with wireless sensor network technology. Keywor ds : arteriovenous shunt, screening sys- tem, hypervolemia, dual-core em- bedded system 1. Introduction Chronic kidney disease is a global public health proble m with high mo rbidity and mo rtal- ity. Treat ment of end-stage renal disease (ESRD) typically involves a kidney transplant or dialysis, which assists damaged kidneys in removing waste and excess fluid fro m the body. In Taiwan, the incidence and prevalent rates of ESRD we re the highest in the wo rld [1], he modia lysis is the most common choice for ESRD patients. Ho w- ever, to perform a he modia lysis, surgeons must create pathological fistulas to provide vascular access routes for treating ESRD. Arteriovenous access (AVA) stenosis is regarded as the pri- ma ry cause of A VA dysfunction in he modia lysis patients and a co mmon occurrence in patients undergoing extended hemodia lysis therapy. According to NKF-DOQI guide lines [2], regula r monitoring and surveillance of A VA function is mandatory. Basically, A VA is a continuous circuit that starts and ends at the heart; it is not simp ly an anastomosis. Moreover, clin ical as- sessments are based on a physical e xa mination of AVA. Th ree physical exa mination steps are look, feel, and listen. Clin ically, A VA flow rates must reach 600– 1000 mL/ min for he modia lysis treatment to be considered efficient. This high-flo w volu me causes vibration of the vas- cular wa ll, which then transmitted to the skin surface, manifesting as a palpable thrill o r aud i- ble bruit [3]. The state of body water is an im- portant factor in the routine hemodialysis (HD) treatment and post-dialysis healthcare. Hyper- volemia is a medica l condition as excess body water in the blood, leading to increases in body sodium (Na) content and a consequent increase in e xt ra-ce llular body water. When dia lysis patients suffer fro m hypertension will increase in we ight and peripheral ede ma in the legs and arms [4-6]. The purpose of this paper is to combine our previous study of audio-phonoangiographic technology in detecting vas cular stenosis with rapid screening system to evaluate dialysis * Corresponding aut hor, Email: kcd56@mail.ncku.edu.tw Advances in Technology Innovation , vol. 2, no. 2, 2017, pp. 46 - 50 47 Copyright © TAETI Fig. 1 Block diagram of the rapid screening system patients’ arteriovenous routes function and their status of excess body fluids: inspecting and inte- grating the blood pressure, body weight, and fistula function work into a rapid screening system, and using the quant izat ion o f fistu la phonoangi- ography pitch to achieve assessing arteriovenous routes. The feasibility of the integrated rapid screening system will be provide an easy opera- tional instrument for efficacious and real time monitoring arteriovenous routes (Fig.1). First, the screening system employed National Instruments (NI) my RIO dual-core embedded microprocessor system integrated with system sensors, software, wireless/wired communications, and output dis- play unit in developing a system which is suitable for rapid screening on hemodialysis units. Second, the screening system is to use audio signal feature extraction algorithms, fuzzy algorithms, and ex- cess body fluid assessment methods on starting begin human status checking. Using vascular echography results as of the narrow criterion to establish the judgment database, and create an alert threshold as indicators of future development, therefore, to achieve the best diagnostic work. Future hoping is developed a complete integrated intelligence system by combining the arteriove- nous fistula signal processing with feature e x- traction with wire less sensor network technol- ogy. 2. Method 2.1. Preliminary Diagnosis and Classification In clinica l research, the degree of narro wing of norma l vessels has been used as an index for the degree of AVA in patients . Exa mination results have been used as a reference to confirm the DOS according to X-ray or sono images . For the measurement segment A site-E site (Fig. 2), the DOS is defined as follows [7-9]. (a) The detection site (b) Sono image of an abnormal blood vessel (c) Spectrum of Short-time Fourier transform (d) Spectrum of Fourier transform Fig. 2 Sono angiography and spectrum analy sis in clinical reasearch Advances in Technology Innovation , vol. 2, no. 2, 2017, pp. 46 - 50 48 Copyright © TAETI %100)1(% 2 2  D d DOS (1) %)%(% postpre DOSDOSDOS  (2) 2.2. Design of fuzzy Petri nets (FPNs)Screening System Based on Fractional-Order Dynamic Error Regarding the surgical outcomes of the 42 p a- tients, the overall pre-PTA DOS% was above 80%, which was regarded as the reference level, whereas the post-PTA DOS% was distributed across three groups (i.e., > 50%, 30%–50%,<30%). No statis- tical significance was observed between the groups at the A sites and V sites (p > 0.05). Therefore, the correlation between DOS% and index  is stronger at the V sites than at the A sites [7]. Ex- ponent regression was used to model the relation- ship between DOS% and index  and between DOS% and index . The prediction model fit a nonlinear curve passing directly through all of the experimental data, as shown in Fig. 3(a). The cor- relation between DOS and index  at the V site can be expressed as: DOS% = 0.1914  exp(0.2333), R 2 = 0.3802 (3) (a) DOS % versus index  at the V site (b) evaluation of the severity of AVA stenosis Fig. 3 Evaluation of the severity of AVA ste- nosis at the V site with DOS % versus index Exponential regression was used to perform a least squares curve fit, which min imized the sum of the squares of the deviations of the e x- perimental data values e xtracted fro m the pre- diction model, thereby obtaining the e xa min a- tion criteria for the predict ion. Fig. 3(b) depicts the range of pre-PTA DOS residuals ( Class I:  < 3, Class II: 3 <  < 5, and Class III:  > 5), which were used as a baseline for eva luating the post-PTA DOS residuals . We also compa red the pre- and post-PTA  indices monthly and de- termined the ideal ranges for evaluating the severity of A VA stenosis in the order of Class III ( < 3), Class II (3 <  < 5), and (Class I:  > 5). (1) A Gaussian me mbe rship function can be parameterized by mean (mean = 0– 6) and standard deviation (1 = 2 = 3 = … = r = 0.3802) by applying Equation (4). ] )( exp[ 2 2 r r r mean      (4) (2) Fig. 4 depicts the Gaussian membership functions for the three classes. Through this approach, we obtained seven me mbership functions r, r = 1, 2, 3, …, 7, with specific ranges denoted as r. The CF of each input in the various ranges is within the range of [0,1]. The FPN can perform fuzzy inference calcu- lations to evaluate the DOS for each propos i- tion specified by a clinical physician. Assume that for the degree of proposition Cm (Class es I–III), m = 1, 2, 3, and place pm is associated with the proposition dm = m(pm), m = 1, 2, 3.. Fig. 4 Gaussian me mbership functions for the three classes of DOS Advances in Technology Innovation , vol. 2, no. 2, 2017, pp. 46 - 50 49 Copyright © TAETI 3. Results and Discussion 3.1 Feasibility Tests with the Proposed Screening System Fig. 5(a) and (b ) show the overall test results ; compared with the DOS% results , the accuracy is 85.71% with six failures. The measurement sites, quantification errors, and undetected ste- nosis could affect the effic iency of the proposed method. The study used at least two 8-second records from the measurement site of 21 patients (i.e ., pre - and post-PA). We determined three DOS classes. The terms UL, L, and VL corre - spond to monotonically decreasing curves that define the degree of ce rtainty. The output of function m may be an arb itrary curve that can be defined as a function that must vary between 0.3679 and 3.000. The place pm can determine the DOS level, and more likely c loses to value 1, 2, or 3 for the goal proposition. If p lace pm is only partially simila r, its value is less than 1 and it gradually decays to 0. Fro m these results, we determined the function of A VS, wh ich can be evaluated using the proposed diagnosis system. (a) Feasibility diagnosis results for the 42 tests (b) degree of stenosis (DOS) versus pm Fig. 5 Feasibility diagnosis results for the 42 tests, and the degree of stenosis (DOS) versus pm Due to variations in the frequency spectra of the various classes and differences a mong the patients, the characteristic frequencies occupy a range of frequency bands, with some characteristic frequencies overlapping or crossing bands. Phys i- cians determined the final degree of certainty as a function of the variance in frequency and magn i- tude. However, obtaining the diagnosis results required an off-line analysis. Traditional Petri nets (PNs) require constant transitions and weighted parameters ( and W), and they encounter dif- ficulty in handling variance in the frequency spectra. PNs were considered appropriate for processing binary data in t rue–false decisions, on–off switching, and automatic control appli- cations [10]. Co mparing FPNs with PNs, the inference rules in the knowledge base of t he rule-based system are both modeled. FPNs use the CF of the fuzzy infe rence rules and weights of the propositions by using binary data and can automatically perform we ighted fuzzy reasoning calculations for analyzing spectral variance. Thus, the proposed rule-based diagnosis system can perform fu zzy infe rence calculat ions in a flexible and intelligent manner. 3.2. Long-Term Examination Using Frequency Parameter-Based Fuzzy Petri Net A 54-year-old fe male patient undergoing hemodia lysis treatment with an A VG (right forearm loop) agreed to part icipate in a long-term e xa mination. Data were collected between June 25, 2011, and July 11, 2012. In a routine monitoring cycle, monthly e xa minations were perfo rmed to evaluate AVG function. Over 3 months of observation, the first characteristic frequency gradually increased fro m 68 to 170 Hz (Fig. 6). On September 6, 2011, the patient presented with a severe AVG occlusion, and a physician confirmed Class II stenosis . Ultrasonic e xa mination indicated a DOS% of 66% at the measurement site, and the patient received PTA treatment. The three characteristic frequencies were 170, 425, and 681 Hz. Fig. 6 Long-term examination at V site from June 25, 2011, to September 6, 2011 Advances in Technology Innovation , vol. 2, no. 2, 2017, pp. 46 - 50 50 Copyright © TAETI 4. Conclusion In an embedded system development env i- ronment, we used four fundamental arithmet ic and logical-reasoning operations to configure the combinational logics or ASICs, and then embedded the intelligent algorithms into a compact chip. The proposed diagnosis system allo ws rule-based configuration and automatic weighted fuzzy reasoning calculations. Fle xib le and intelligent algorith ms require no iteration for updating system we ights. Therefore, this system can handle comple x configuration designs and is appropriate for the short design cycle of prot o- type imp le mentation, testing, debugging, and modification. Currently, graphical user interfaces of Windows-based applications enable reprogramming and have flexible architectures that facilitate the rapid development of customized ASICs. The proposed FPN algorithm can be implemented using four fun- damental operations with specific data structures. Therefore, it can also be implemented with the four fundamental operations. By combining an FPN and logical operation functions, the proposed diagnosis system provides a promising means for implement- ing a portable monitor for AVS evaluation in home care. Acknowledgement This study is supported in part by the re- search grant of Kaohsiung Veterans Genera l Hospital (VGHKS105-070) and the Ministry of Science and Technology, Taiwan (MOST), under contract number: MOST 105-2218-E-075B -001. References [1] A. J. Collins et al., “ USRDS 2012 annual data report: atlas of chronic kidney disease and end-stage renal disease in the United States ,” American J. Kindeny Diseases, vol. 59, pp. 342, 2012. [2] N. K. Foundation, “KDOQI clinical practice guidelines and clinical practice reco mmendations for 2006 updates: he modialysis adequacy, peritoneal dia lysis adequacy and vascular access,” Am. J. Kidney Dis., vol. 48, pp. 1-322, 2006. [3] F. Loth, P. F. Fischer, and H. S. Bassiouny, “Blood flow in end-to-side anastomoses,” Annu. Rev. Fluid Mech., vol. 40, pp. 367-393, 2008. [4] P. W. Cha mney, K. Matthias, C. Rode, W. Kle inekofort, and V. W ize mann, “A new technique for establishing dry we ight in hemodia lysis patients via whole body bi- oimpedance,” Kidney International, vol. 61, pp. 2250-2258, 2002. [5] A. H. T za malou kas, G. H. Murata, D. J. Vanderjagt, K. S. Serv illa, and R. H. Glew, “Body composition evaluation in peritoneal dia lysis pat ients us ing ant hro po metric formulas estimat ing body water,” Advances in Pe ritoneal Dialysis, vol. 19, pp. 212-216, 2003. [6] R. Agarwal, “Hypervolemia is associated with increased mortality among hemodialysis patients,” Hypertension, vol. 56, pp. 512-517, 2010. [7] Y. C. Du, W. L. Chen, C. H. Lin, C. D. Kan, and M. J. Wu, “Residual stenosis estimation of arteriovenous grafts using a dual channel honoangiography with fract iona l-o rder features,” IEEE Journal of Bio medica l and Health Informat ics, vol. 18, no. 2, pp. 703-713, 2014. [8] W. L. Chen, C. H. Lin, T. Chen, P. J. Chen, and C. D. Kan, “Phono angiography with a fractional order chaotic system - A new and easy algorithm in analyzing residual arterio- venous,” Medical & Biological Eng & Computing, vol. 51, pp. 1011-1019, 2013. [9] W. L. Chen, C. H. Lin, T. Chen, P. J. Chen, and C. D. Kan, “Stenosis detection using burg method with autoregressive model for hemo- dialysis patients,” Journal of Medical and Bio medical Engineering, vol. 33, no. 4, pp. 356-362, 2013. [10] R. S. Lees and C. F. De wey, “Phono angi- ography: a new noninvasive diagnostic method for studying arterial d isease,” Pro- ceedings of the National Academy of Sci- ences, vol. 67, no. 2, pp. 935-942, 1970.