Date on Master's Thesis/Doctoral Dissertation
8-2024
Document Type
Master's Thesis
Degree Name
M. Eng.
Department
Chemical Engineering
Degree Program
JB Speed School of Engineering
Committee Chair
Berson, Eric
Committee Member
Jaeger, Vance
Committee Member
Hill, Bradford
Author's Keywords
blood age; FFR; TKE; neural network; stenosis
Abstract
In this study, contour fields of mean age and turbulence kinetic energy (TKE) in the post-stenotic region of model arteries were analyzed to explore their correlation with FFR. While both high and low FFR cases exhibited similarities with an inner low-age region and an outer high-age region, low FFR cases displayed asymmetry in their low-age regions due to eccentric stenosis. Moreover, gradients of age across the artery diameter were more pronounced for low FFR cases. Additionally, low FFR cases demonstrated significantly higher TKE in the post-stenotic blood jet stream compared to high FFR cases that are reflected in low-age boundary deformations.
Blood age distributions in post-stenotic regions were inputted into a convolutional graphical neural network to predict whether high or low FFR was present for a given case. Despite a small sample size used for training and evaluation, the constructed network achieved an accuracy of 75%, an F1 score of 83%, and a specificity and recall of 86% on the evaluation dataset, suggesting successful pattern recognition. These findings imply that the age distribution patterns observed in the contours were not likely coincidental. Although preliminary results of the network indicate strong potential for future diagnoses, larger training and evaluation datasets are necessary to make more definitive statements about its performance.
Recommended Citation
LaRoche, Michael, "Analysis of the relationship between flow patterns in the post-stenotic region of model arteries and stenosis severity." (2024). Electronic Theses and Dissertations. Paper 4411.
Retrieved from https://ir.library.louisville.edu/etd/4411
