Preliminary Experimental Validation
To further validate the proposed fuzzy control method, a preliminary in vitro experiment was also performed. The experimental rig is shown in Figure 8A. It contains a pressure sensor (CEMPX-2; Nokisens Inc., Kunshan, China) and a flow meter (T402; Transonic System, Ithaca, NY). The pressure sensor obtains the pump’s outlet pressure, which is regarded as the aortic pressure. A centrifugal blood pump designed by ourselves is used as the control object. The vascular resistance is reproduced by a throttle valve. The aortic pressure is acquired and sent to the computer through a data acquisition card (PCI-1712, Advantech, Taipei, Taiwan). The fuzzy controller is realized in LabVIEW software (National Instruments, Austin, TX).
The effects of the pulse pressure–enhancing controller under the condition of heart failure are shown in Figure 5, which also includes the constant speed control results. Figure 5 shows the aortic pressure in several cardiac cycles. Both the pulse pressure–enhancing control and the constant speed control elevate the aortic pressure to a mean value of 100 mm Hg, which corresponds to a mean circulatory flow rate of approximately 100 ml/s. However, the constant speed control reduces the pulse pressure to an extremely low level, even much less than the heart failure condition without an LVAD. Compared with the constant speed control, it is clear that the pulse pressure is enhanced to approximately 20 mm Hg by the pulse pressure–enhancing control. Under this circumstance, the pump speed profile is rectangular wave, as shown in Figure 5B. It is noted here that the heart period is also changed by the baroreflex system because of the change of aortic pressure after controlling. In addition, the left ventricular pressure–volume (P-V) loops are displayed in Figure 5C. Either of the control methods can unload the failing left ventricle, but the constant speed control relieves the ventricle more. Stroke work under pulse-enhancing control is approximately 56.5% of that in the heart failure condition without an LVAD compared with a value of 26.4% under constant speed control.
The controller was tested under various physiologic disturbances, which were more rigorous than actual disturbances. Good results were yielded, which are depicted in Figure 6. It is clear that the controller could withstand these disturbances, thus showing good robustness. When Rsvr increases from 0.9 to 1.2 mm Hg·s/ml at time 100 s, MPI deviates from the set point but quickly stabilizes back to the desired value, while PPI decreases to approximately 5 and stays there because of regurgitation prevention, thus limiting the speed variation amplitude. After Rsvr returns to its initial value at 150 s and further decreases to 0.6 mm Hg·s/ml at 200 s, PPI increases to 6 again and then increases past 6 and remains unchanged at this level. Cao changes, from 2.0 to 2.5 mm Hg/ml and then to 1.5 ml/mm Hg during 300–450 s, affect PPI a lot while only having little influence on MPI. Elv,s changes in a similar way (among 0.6, 0.3, and 0.1 mm Hg/ml) from 500 s, and the controller still performs well. During all the simulation times, once the backflow is happening in the pump, the regurgitation prevention method takes effect and pulls the pump flow back to positive, as shown in Figure 6E. Figure 6F shows the heart period variation, which has a similar trend to MPI because of the baroreflex system.
Slight exercise in a heart failure patient was simulated by linearly changing the Rsvr (decreased linearly by 0.3 mm Hg·s/ml), Cao (increased linearly by 0.3 ml/mm Hg), and
(increased linearly by 0.3 mm Hg/ml) simultaneously. Figure 7 shows the responses to this physical activity. The controller still performs well during the transition from rest to exercise and back to rest. When in exercise, left ventricular pressure (LVP) decreases to low levels because more blood is pumped out to meet the metabolic demand. The pump flow shown in Figure 7D indicates that regurgitation is less likely to take place during exercise.
Figure 8B shows the results of the preliminary experiment. Similar to the numerical simulation, the controller reaches the set points very well. The mean pressure of the pump’s outlet is kept at 100 mm Hg to provide a mean flow of approximately 5 L/min, and the pump alone generates a pulse pressure of approximately 20 mm Hg. The experiment demonstrates that the dynamic characteristic of a general blood pump is sufficient to track a speed modulation in one cardiac cycle. The controller is feasible in practical use. More experiments on a mock circulatory system or in animal trials need to be implemented to further validate the controller.
The controller proposed in this study is successful in enhancing the pulse pressure compared with the constant speed controller, but at the cost of less unloading of the ventricle. We can see this from the P-V loops in Figure 5. However, because excessive unloading of the left ventricle might lead to cardiac disuse atrophy,19 it is reasonable and acceptable to use 56.5% of the stroke work during no-LVAD-assisting heart failure for unloading. On the contrary, as shown in Figure 6B, PPI stays in the range 6–14 in some cases, as expected, which is due to the particular setting of the zero error membership function in Figure 4B. In this manner, the controller will not decrease the pulse pressure if the ventricle recovers and increases pulsation, unless the pulse pressure becomes more than 50 mm Hg as a result of other diseases such as arteriosclerosis.
Another notable aspect is that the LVP may decrease low values when the
reduces, as shown in Figure 6C during 200–250 s. A similar situation occurs when a patient is doing exercise (Figure 7), mainly because of the reduction in resistance. If the
decreases to abnormal levels because of drugs or over exercise, a suction event might happen. Thus, suction detection is needed. The pump flow signal acquired for regurgitation prevention can also be used to develop a suction detection system.20 It is worth noting that suction under this control method is a little different from traditional constant speed control. Suction occurs intermittently because of speed variation and only for a short time in every cardiac cycle. The effect of this phenomenon on the heart is not clear and needs to be further investigated.
There are two necessary signals in this study, the aortic pressure used to extract the indices and the pump flow used to prevent regurgitation. They can either be obtained by sensors implanted in the blood pump or estimated noninvasively using intrinsic pump parameters, such as current, voltage, and speed.9,10,21 However, it is more accurate to use sensors than estimation. Therefore, sensor is a better choice, especially during short-term transition therapy. Estimation might be better for long-term use. Take pressure sensors for example, although some of the long-term implantable pressure sensors are already available,22,23 reliability must be improved. Sensor drift influences MPI more than PPI. Baseline drift has almost no effect on PPI. However, if sensors are picked for long-term use, it is still necessary to perform regular calibration in a medical center.
In this study, the rectangular wave is adopted to modulate the pump speed because it increases the pulse pressure most compared with other basic waveforms such as a sine signal.24 The phase shift and high pump speed period defined in Figure 3B affect pulsation enhancement, which is depicted in Figure 9,AandB. The results are obtained when the pump works under maximum speed variation amplitude that prevents regurgitation. From Figure 9A, we can see that the pulse pressure is enhanced the most when the phase shift is zero, which fits our objective. If other considerations, such as a specific ventricular unloading are included, more investigations are required to find the optimal phase shift. The pulse pressure has an inverse relation with the high pump speed period, as shown in Figure 9B. This is because a shorter high pump speed period means that the mean pump speed must increase under the same desired mean aortic pressure, which can increase the maximum speed variation amplitude before regurgitation. However, too short of a high pump speed period is not achievable in practice because of the limited pump response. Similarly, a higher desired mean aortic pressure also increases the mean pump speed and thus leads to a higher pulse pressure the controller can generate before regurgitation, as shown in Figure 9C. Therefore, an appropriate increase in the MPI set point is desirable because it makes the controller easier to achieve the minimum pulse pressure of 20 mm Hg.
Theoretically, a permanent magnet motor is rapid enough to let the pump accelerate and decelerate during every single cardiac cycle because its mechanical and electrical time constants are much less than the heart period.25 Moreover, within the power limitation, high speed accuracy can be achieved with feedback control of the motor. As a result, the speed output of the controller is able to be realized in a real blood pump, which has been demonstrated by the preliminary experiment and some other experiments.11,16
Pulsatile control of continuous flow blood pumps is a new topic and attracts more and more attention worldwide. More investigations support the point that pulsatile perfusion is better.5,26 In continuous flow blood pump therapy, current clinical investigations also suggest some negative effects such as arteriovenous malformations and bleeding events related to diminished pulse pressure,27 implying it will be beneficial to control pulsation of RBPs even for short-term transition therapy. Unlike other pulsatile control strategies using the inherent pulsatility in the native ventricle,7,11 the controller proposed by us actively regulates the pump speed to produce extra pulsation. The controller may also work well in advanced heart failure patients whose hearts cannot generate enough pulsation. Except for pulsation, sufficient perfusion is achieved by maintaining the mean aortic pressure at 100 mm Hg. Furthermore, compared with the previous investigations that focus on the physiologic effects of typical fixed speed modulations,16,24 the speed modulation in our study is autocontrolled to fit various physiologic conditions. It is noted that this control needs ECG signal to synchronize the pump with the natural heart. If arrhythmia exists or the heart is totally removed, a modulation waveform with a fixed cycle may be used instead. Another two concerns of speed modulation are whether there will be increased wear to the bearings and additional shear stress to the blood. It is likely to have increased wear, but the extent of the increase is not explicit and needs further investigations. Fortunately, the latest generation of blood pumps is usually suspended without contact bearings, so there is no need to worry about such a problem. As for shear stress, a previous group has demonstrated that there was no excessive hemolysis due to the speed alternation.28 However, more investigations are needed to support the conclusion.
Fuzzy control is believed to be superior in control applications in which the mathematic model of the cardiovascular system is limited and uncertain. Other fuzzy control studies of blood pumps have also shown advantages.29,30 The controller can be easily transplanted between multiple patients with different kinds of blood pumps. Both the simulation with the cardiovascular system and the preliminary experiment in vitro demonstrated the performance of this fuzzy controller, as shown in the results.
A novel approach to control an RBP in advanced heart failure patients has been proposed in this article. It can provide additional pulsation while maintaining the mean aortic pressure. The pulse pressure is enhanced to a normal physiologic minimum value of 20 mm Hg by means of continuous regulation of the pump speed according to a modulating signal, which is synchronous with the cardiac cycle. The mean aortic pressure of 100 mm Hg is also achieved to provide sufficient flow. Perfusion will benefit from the adequate flow and enhanced pulsation. Fuzzy logical inference is used in this research, and it is very simple and easy to be applied in the field of blood pump control. Numerical simulations under various physiologic disturbances have been performed to demonstrate the effectiveness and robustness of the algorithm, and satisfying results are delivered. A preliminary in vitro experiment has also been implemented to prove the feasibility of the controller in practice. Future tests on a mock circulatory system or even in vivo experiments are needed to further verify the reliability of the control strategy.
The authors are grateful to National Natural Science Foundation of China (Grant No. 51275461), the Science Fund for Creative Research Groups of the National Natural Science Foundation of China (Grant No. 51221004), and Zhejiang Provincial Natural Science Foundation of China (Grant No. Z1110189).
1. Ohuchi K, Kikugawa D, Takahashi K, et al. Control strategy for rotary blood pumps. Artif Organs. 2001;25:366–370
2. Giridharan GA, Skliar M, Olsen DB, Pantalos GM. Modeling and control of a brushless DC axial flow ventricular assist device. ASAIO J. 2002;48:272–289
3. Giridharan GA, Pantalos GM, Gillars KJ, Koenig SC, Skliar M. Physiologic control of rotary blood pumps: an in vitro
study. ASAIO J. 2004;50:403–409
4. Gwak KW. Application of extremum seeking control to turbodynamic blood pumps. ASAIO J. 2007;53:403–409
5. Alkan T, Akçevin A, Undar A, Türkoğlu H, Paker T, Aytaç A. Benefits of pulsatile perfusion on vital organ recovery during and after pediatric open heart surgery. ASAIO J. 2007;53:651–654
6. Hornick P, Taylor K. Pulsatile and nonpulsatile perfusion: the continuing controversy. J Cardiothorac Vasc Anesth. 1997;11:310–315
7. Choi S, Antaki J, Boston R, Thomas D. A sensorless approach to control of a turbodynamic left ventricular assist system. IEEE Transactions on Control Systems Technology. 2001;9:473–482
8. Shi Y, Korakianitis T. Numerical simulation of cardiovascular dynamics with left heart failure and in-series pulsatile ventricular assist device. Artif Organs. 2006;30:929–948
9. Lim E, Alomari AH, Savkin AV, et al. A method for control of an implantable rotary blood pump for heart failure patients using noninvasive measurements. Artif Organs. 2011;35:E174–E180
10. Giridharan GA, Skliar M. Physiological control of blood pumps using intrinsic pump parameters: A computer simulation study. Artif Organs. 2006;30:301–307
11. Gao B, Chang Y, Gu K, Zeng Y, Liu Y. A pulsatile control algorithm of continuous-flow pump for heart recovery. ASAIO J. 2012;58:343–352
12. Vollkron M, Schima H, Huber L, Wieselthaler G. Interaction of the cardiovascular system with an implanted rotary assist device: Simulation study with a refined computer model. Artif Organs. 2002;26:349–359
13. Ursino M. Interaction between carotid baroregulation and the pulsating heart: A mathematical model. Am J Physiol. 1998;275(5 pt 2):H1733–H1747
14. Chen S, Ferreira A, Simaan MA, Antaki JF. A mathematic model of a cardiovascular system regulated by the baroreflex. American Control Conference
; 2006;2006 IEEE:p. 6
15. Choi S, Boston J, Thomas D, Antaki JF. Modeling and identification of an axial flow blood pump. American Control Conference, 1997 Proceedings of the 1997
; 1997;1997 IEEE:3714–3715
16. Pirbodaghi T, Weber A, Axiak S, Carrel T, Vandenberghe S. Asymmetric speed modulation of a rotary blood pump affects ventricular unloading. Eur J Cardiothorac Surg. 2013;43:383–388
17. Passino KM, Yurkovich S. Fuzzy control 1998 Citeseer
18. Kingwell BA, Berry KL, Cameron JD, Jennings GL, Dart AM. Arterial compliance increases after moderate-intensity cycling. Am J Physiol. 1997;273(5 pt 2):H2186–H2191
19. Kinoshita M, Takano H, Taenaka Y, et al. Cardiac disuse atrophy during LVAD pumping. ASAIO Trans. 1988;34:208–212
20. Vollkron M, Schima H, Huber L, Benkowski R, Morello G, Wieselthaler G. Development of a suction detection system for axial blood pumps. Artif Organs. 2004;28:709–716
21. AlOmari AH, Savkin AV, Karantonis DM, Lim E, Lovell NH. Non-invasive estimation of pulsatile flow and differential pressure in an implantable rotary blood pump for heart failure patients. Physiol Meas. 2009;30:371–386
22. Cong P, Young DJ, Hoit B, Ko WH. Novel long-term implantable blood pressure monitoring system with reduced baseline drift. Conf Proc IEEE Eng Med Biol Soc. 2006;1:1854–1857
23. Bullister E, Reich S, D’Entremont P, Silverman N, Sluetz J. A blood pressure sensor for long-term implantation. Artif Organs. 2001;25:376–379
24. Pirbodaghi T, Axiak S, Weber A, Gempp T, Vandenberghe S. Pulsatile control of rotary blood pumps: Does the modulation waveform matter? J Thorac Cardiovasc Surg. 2012;144:970–977
25. Xu L, Fu M. Computer modeling of interactions of an electric motor, circulatory system, and rotary blood pump. ASAIO J. 2000;46:604–611
26. Undar A, Masai T, Yang SQ, et al. Pulsatile perfusion improves regional myocardial blood flow during and after hypothermic cardiopulmonary bypass in a neonatal piglet model. ASAIO J. 2002;48:90–95
27. Soucy KG, Koenig SC, Giridharan GA, Sobieski MA, Slaughter MS. Rotary pumps and diminished pulsatility: Do we need a pulse? ASAIO J. 2013;59:355–366
28. Tayama E, Nakazawa T, Takami Y, et al. The hemolysis test of the Gyro C1E3 pump in pulsatile mode. Artif Organs. 1997;21:675–679
29. Fu M, Xu L. Computer simulation of sensorless fuzzy control of a rotary blood pump to assure normal physiology. ASAIO J. 2000;46:273–278
30. Casas F, Ahmed N, Reeves A. Minimal sensor count approach to fuzzy logic rotary blood pump flow control. ASAIO J. 2007;53:140–146
Keywords:Copyright © 2014 by the American Society for Artificial Internal Organs
physiologic control; fuzzy logic; pulse pressure; rotary blood pump; ventricular assist device