| Heart transplantation is still the golden standard for the treatment of advanced heart failure.However,the donor hearts is largely insufficient to cater for the demand.Therefore,in the past fifty years,various blood pumps have been developed for partially or fully replacing the function of the natural heart.During the assist period,the non-physiological flow patterns in the blood pump,such as high shear stress,long exposure time,recirculation and flow stagnation,may cause the damage of blood cells,including the rupture or mechanical property change of red blood cells(i.e.,hemolysis),the activation of platelets and white blood cells,the increased concentration of inflammatory mediators,embolism and thrombosis in blood pumps.The blood damge would lead to various complications and consequently endanger the patient‘s life.To reduce the blood damage related complications and augment the patients survival rate,optimizing the inner flow field of the blood pumps has become one of the most significant issues in the design of blood pumps.However,previous investigations associated with blood damage mainly focus on hemolysis or on thrombosis,and most of them are improvements other than automatic optimization.Due to the simulation complexity of pulsatile blood pumps,there are few CFD-based investigations on their blood damge.Since the operating conditions also have influences on the blood damage of blood pumps,it is necessary to comprehensively study their influences on the blood damge.In the paper,for the unresolved issues associated with the blood damge,the flow simulation of a pulsatile blood pump was conducted by fluid-structure interaction technology,and the blood damge was predicted based on the simulaiton results;The geometric parameters of the blood chamber were optimized to reduce its blood damge;By studying the influences of different operating conditions on the blood damge,the operating conditions that reduce the blood damge can be obtained;A sensorless predictive model was constructed,and it can predict the output parameters and blood damage indices using the driving parameters of the blood pump.The main contents of this paper are as follows:1.The flow field simulation of the pulsatile blood pump was conducted by fluid-structure interaction,including the motion and deformation of the sac and the rotary motion of the valves.The simulations of three different valve modeling(no valve,fixed valve and rotary vlave)were compared,and the corresponding applicable occasions were analyzed with PIV validation.The hemolysis index and two thrombosis indices(including platelet activation index and platelet deposition index)were predicted based on the simulation results,and the hemolysis index was validated by in vitro experiment,which provided the foundation for the optimization of the pulsatile blood pump.2.A multi-objective optimization framework integrating FSI,response surface and NSGA-II was proposed,and it was applied to optimize three blood chamber models(Model I,Model II and Model III),using three blood damage indices as the goal functions.And the clinical applications of the three models were also discussed.3.The influences of different operating conditions(including plate motion profile,pulsatile rate,stroke volume and assist mode)on the blood damage(including hemolysis,platelet activation and platelet deposition)of pulsatile blood pumps were investigated.The results indicated that cosine motion profile(compared to sine and polynomial),higher ejection fraction,higher pulsatile rate and counter-pulsation(compared to co-pulsation)can decrease the potential of platelet deposition whereas increase the potential of hemolysis and platelet activation,and vice versa.Additionally,the results also suggested that the three blood damage indices cannot be reduced simultaneously.4.A neural networks-based prediction method for the output parameters of the pulsatile blood pump was built,which can predict the average output pressure,the average output flow rate and the blood damage indices(including IH,Dpl and Total TSP)by the driving parameters(including motor driving frequency,average driving current and pulsatile rate).The predictive model was validated in the mock circulatory system.The average percent predictive error of the proposed method is lower than the reported methods.Using neural networks in the blood damage prediction is reported for the first time by this paper. |
