Handmade trileaflet valve design and validation for patch-valved conduit reconstruction using generalized regression machine learning model

Chung Dann Kan, Jieh Neng Wang, Chia Hung Lin, Wei Ling Chen, Pong Jeu Lu, Ming Yao Chan, Jui Te Wu

Research output: Contribution to journalArticlepeer-review

Abstract

Pulmonary valve diseases include the different degrees of aortic stenosis or congenital defects in children or adults. Valve repair or replacement surgery is commonly performed to relieve valvular dysfunction and improve the significant flow regurgitation in the aortic valve and the pulmonary valve. However, commercial valve stents and valved conduits are sometimes not available for children or patients with special conditions. The handmade trileaflet valve design has been used with different range of diameters for patch-valved conduit reconstruction. Thus, we propose a multiple regression model, as a generalized regression neural network (GRNN), to determine the optimal trileaflet parameters, including the width, length, and upper lower curved structure. Through computed tomography pulmonary angiography, while the diameter of the main pulmonary artery is determined, a leaflet template can be rapidly sketched and made. Using an experimental pulmonary circulation loop system, the efficacy of the valved conduit can be validated using the regurgitation fraction method. In contrast to commercial valve stents, experimental results indicate that the handmade trileaflet valve can also improve severe pulmonary regurgitations.

Original languageEnglish
Pages (from-to)605-620
Number of pages16
JournalTechnology and Health Care
Volume26
Issue number4
DOIs
Publication statusPublished - 2018 Jan 1

All Science Journal Classification (ASJC) codes

  • Biophysics
  • Bioengineering
  • Biomaterials
  • Information Systems
  • Biomedical Engineering
  • Health Informatics

Fingerprint Dive into the research topics of 'Handmade trileaflet valve design and validation for patch-valved conduit reconstruction using generalized regression machine learning model'. Together they form a unique fingerprint.

Cite this