Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Post-processing of coronary and myocardial spatial data

Jay Aodh Mackenzie, Megan Jeanne Miller, Nicholas Hill, Mette Olufsen

TL;DR

The paper tackles the challenge of constructing computational haemodynamics domains from a partial coronary arterial graph to enable accurate 1D perfusion simulations of the myocardium. It presents a data-processing pipeline that cleans the vascular graph, assigns generation levels, and extracts principal pathways through radius-, density-, and Strahler-based filters, balancing fidelity with computational cost via information-density metrics. The ventricular subdivision is then performed by linking downstream arterial regions to LV tissue, creating disjoint subdomains and comparing them to the AHA regional framework to facilitate clinical communication. The work demonstrates robust, scalable methods applicable to porcine data and other similarly structured vascular networks, with implications for patient-specific modelling and broader organ perfusion studies.

Abstract

Numerical simulations of real-world phenomena require a computational scheme and a computational domain. In the context of haemodynamics, the computational domain is the blood vessel network through which blood flows. Such networks contain millions of vessels that are joined in series and in parallel. It is computationally unfeasible to explicitly simulate blood flow throughout the network. From a single porcine left coronary arterial tree, we develop a data pipeline to obtain computational domains for haemodynamic simulations in the myocardium from a graph representing a partial coronary arterial tree. In addition, we develop a method to ascertain which subregions of the left-ventricular wall are more likely to be perfused via a given artery, using a comparison with the American Heart Association division of the left ventricle for validation.

Post-processing of coronary and myocardial spatial data

TL;DR

The paper tackles the challenge of constructing computational haemodynamics domains from a partial coronary arterial graph to enable accurate 1D perfusion simulations of the myocardium. It presents a data-processing pipeline that cleans the vascular graph, assigns generation levels, and extracts principal pathways through radius-, density-, and Strahler-based filters, balancing fidelity with computational cost via information-density metrics. The ventricular subdivision is then performed by linking downstream arterial regions to LV tissue, creating disjoint subdomains and comparing them to the AHA regional framework to facilitate clinical communication. The work demonstrates robust, scalable methods applicable to porcine data and other similarly structured vascular networks, with implications for patient-specific modelling and broader organ perfusion studies.

Abstract

Numerical simulations of real-world phenomena require a computational scheme and a computational domain. In the context of haemodynamics, the computational domain is the blood vessel network through which blood flows. Such networks contain millions of vessels that are joined in series and in parallel. It is computationally unfeasible to explicitly simulate blood flow throughout the network. From a single porcine left coronary arterial tree, we develop a data pipeline to obtain computational domains for haemodynamic simulations in the myocardium from a graph representing a partial coronary arterial tree. In addition, we develop a method to ascertain which subregions of the left-ventricular wall are more likely to be perfused via a given artery, using a comparison with the American Heart Association division of the left ventricle for validation.
Paper Structure (25 sections, 4 equations, 25 figures)

This paper contains 25 sections, 4 equations, 25 figures.

Table of Contents

  1. Introduction
  2. Materials & Methods
  3. Error Correction and Graph Simplification
  4. Generation Assignation
  5. Obtaining Principal Pathways
  6. Setting a Maximum Number of Generations
  7. Radius-Based Metrics
  8. Removing Disconnected Segments
  9. Segment Concatenation
  10. Further Length Based Filtration
  11. Strahler Ordering
  12. Ventricular Data
  13. The Septum
  14. Initial subdivision
  15. Creating Disjoint Subdomains
  16. ...and 10 more sections

Figures (25)

  • Figure 1: A lateral view of the segments of the vascular tree. Line colour demarcates the segments. The root node is marked with a red plus.
  • Figure 2: Coronal projection of the segments are sorted into generations. Line weight and colour signify generation affiliation. (a) All segments are sorted into generations without modification. (b) 30-fold magnification of the initial junction in the tree shown in (a); spatial nodes are highlighted with circles and the short segment is clear to see. (c) Pseudotrifurcations are removed and the resulting tree is shown with the same view as (a). (d) as in panel (b), but in the tree without pseudo-trifurcations.
  • Figure 3: The panels show (a) the number of vessels, (b) the total volume of those segments, and (c) the mean volume of a segment in each of the 32 generations. The mean volume of a segment in each generation is the quotient of the total volume of a single generation by the number of segments in that generation.
  • Figure 4: The evolution of a tree as it passes through some of the filters discussed here; line weight and colour indicate generation affiliation; the Mercator projection preserves the tree structure but not segment length. (a) Generations 0 -- 6 of the tree without pseudotrifurcations. (b) Segments from (a) that additionally have mean radius of at least 0.6 mm. (c) The segments from (b) have been joined in series. (d) Short terminal segments from (c) have been removed and the series join has been repeated.
  • Figure 5: Radius filtered and pruned trees with a threshold set at 0.6 mm. The largest connected subtrees are shown in solid black and disconnected segments in cyan. (a) The mean radius condition is applied; there are 47 connected segments and 21 disconnected. (b) A proportion threshold is set to 0.8 and applied; there are 26 connected segments and 21 disconnected. (c) A single point threshold is applied to yield a connected tree of 75 connected segments and 53 disconnected.
  • ...and 20 more figures