Morphological Analusis Of The Left Ventricular Eendocardial Surface Using A Bag-Of-Features Descriptor

TL;DR

This paper introduces a novel shape analysis method using a Bag-of-Features descriptor to analyze the morphology of the left ventricular endocardial surface from high-resolution CT images, aiding in the detection of coronary artery disease.

Contribution

It proposes a new shape analysis approach combining shape descriptors and Bag-of-Features for localizing and quantifying coronary artery disease from cardiac imaging.

Findings

Successful identification and localization of CAD-related morphological changes.

Quantitative correlation between endocardial surface alterations and coronary stenosis severity.

Potential clinical impact in non-invasive CAD assessment.

Abstract

The limitations of conventional imaging techniques have hitherto precluded a thorough and formal investigation of the complex morphology of the left ventricular (LV) endocardial surface and its relation to the severity of Coronary Artery Disease (CAD). Recent developments in high-resolution Multirow-Detector Computed Tomography (MDCT) scanner technology have enabled the imaging of LV endocardial surface morphology in a single heart beat. Analysis of high-resolution Computed Tomography (CT) images from a 320-MDCT scanner allows the study of the relationship between percent Diameter Stenosis (DS) of the major coronary arteries and localization of the cardiac segments affected by coronary arterial stenosis. In this paper a novel approach for the analysis using a combination of rigid transformation-invariant shape descriptors and a more generalized isometry-invariant Bag-of-Features (BoF) descriptor, is proposed and implemented. The proposed approach is shown to be successful in identifying, localizing and quantifying the incidence and extent of CAD and thus, is seen to have a potentially significant clinical impact. Specifically, the association between the incidence and extent of CAD, determined via the percent DS measurements of the major coronary arteries, and the alterations in the endocardial surface morphology is formally quantified. A multivariate regression test performed on a strict leave-one-out basis are shown to exhibit a distinct pattern in terms of the correlation coefficient within the cardiac segments where the incidence of coronary arterial stenosis is localized.