P65: Dual-modality Volume Estimation And Its Impact On A Preload-Sensitive Controller

ASAIO Journal, vol. 68, pp. 130

Abstract

Purpose of study: In severe heart failure, ventricular assist devices (VADs) are critical to extend patients’ lives. Currently, VADs run at constant speed, whereas physiologic control could improve treatment. A very robust choice for such control is the left ventricular volume (LVV). However, continuously monitoring LVV is challenging, as the measurement modalities are only calibrated during implantation, and accuracy decreases during subsequent geometric remodeling of the heart. As a result, LVV control has not been realized. Methods: We integrated ultrasound and bioimpedance modalities on a cannula of an apical VAD. Both modalities were combined by obtaining a least-squares estimate. In-vitro, we quantified LVV measurement accuracies of each individual modality, and the combined version over a wide volume range (140-420 mL). These accuracies were fed insilico into a preload sensitive controller (PLSC) interacting with a cardiovascular computational model. Simulating a dilated heart (333 mL) that remodels to physiologic volumes (133 mL), for example, within 3 months after implantation, we report the effect of individual and combined LVV estimation on controlled cardiac output (CO). Results: The in-vitro experiments showed that the individual modalities had estimation errors up to -27% to -90% in the remodeled heart, which was improved to -10% for the dual-modality approach. This resulted in improved behavior of the controller as shown in Figure 1. The PLSC generated physiological CO with all modalities (top) shortly after simulated implantation due to controller calibration. In contrast, in the remodeled heart, the low accuracy of individual modalities (bioimpedance and ultrasound) resulted in low CO (bottom), while the dual-modality approach generated sufficient CO. Overall, the controller is less aggressive with large estimation errors and tracks physiological behavior with the dual-modal approach. Conclusion: The dual-modality approach shows low error in LVV estimation even if LVV changes. Thus, a dual-modality approach may solve the calibration conundrum of the PLSC while the heart remodels in the first three months post-implantation.

Authors 5

1. Leonie Korn Aachen Medical Information Technology (MedIT)

   RWTH Aachen University

   Affiliation as printed

   Medical Information Technology, RWTH Aachen University, Aachen, Germany

2. Steffen K. Leonhardt Aachen Medical Information Technology (MedIT)

   RWTH Aachen University

   Affiliation as printed

   Medical Information Technology, RWTH Aachen University, Aachen, Germany

3. Melanie N. Zeilinger

   ETH Zurich · Institute for Dynamic Systems and Control

   Affiliation as printed

   Institute for Dynamic Systems and Control, ETH Zurich, Zurich, Switzerland

4. Marianne Schmid Daners

   ETH Zurich

   Affiliation as printed

   Product Development Group Zurich, ETH Zurich, Zurich, Switzerland

5. Seraina Anne Dual

   KTH Royal Institute of Technology

   Affiliation as printed

   Department of Biomedical Engineering and Health Systems, KTH Royal Institute of Technology, Stockholm, Sweden