Use this URL to cite or link to this record in EThOS:
Title: The quantification of cardiac function
Author: Cole, Graham David
ISNI:       0000 0004 7969 8164
Awarding Body: Imperial College London
Current Institution: Imperial College London
Date of Award: 2017
Availability of Full Text:
Access from EThOS:
Access from Institution:
Quantification of cardiac function is a crucial part of diagnosis, prognosis and ongoing management for many patients with heart disease. Clinical examination is an insensitive marker of cardiac dysfunction, resulting in echocardiography being an in-demand gold-standard test. In recent years, there has been a revolution in point-of-care echocardiography, with the availability of affordable equipment with limited functionality capable of obtaining 2-dimensional pictures and limited measurements. This equipment is often used by non-specialists who make visual gradings or estimates of cardiac function. In this thesis I showed that this approach is potentially flawed, in that operators find it difficult to agree with themselves (and consequently with each other) when reviewing the same images. Even when used as a "gatekeeper" test to differentiate normal from abnormal in triage for further testing, this thesis shows that operators frequently disagree with their original judgement when seeing the same images again on the question of normal versus abnormal. I developed software to allow humans to make bias-resistant replicate measurements from Doppler traces blinded to their previous results and used this to investigate how operators make measurements of cardiac function and aortic stenosis. I show that the mechanism of disagreement in tracing velocity time integral is the difficulty in tracing steep slopes at the beginning and end of the trace. I also show that one of the reasons why dimensionless index is useful is that it is the ratio of correlated errors: operators tend to be systematic "over-readers" or "under-readers". Tissue Doppler imaging can be used to assess cardiac function, and because it encodes systolic and diastolic information efficiently, is ideal for automated image analysis. In this thesis I show that current clinical protocols capture most of the variability that arises on a beat to beat basis, but that subtle differences in optimal probe placement (unrecognised by an operator) are an unpredictable and potentially significant source of variability. Because beat to beat variation arises early on, extended multi-beat analysis is capable of generating high-precision measurements, which may be of value in research. I go on to develop an image analysis algorithm to extract measurements from tissue Doppler traces, testing it against human operators. I show that the system can make measurements of systolic and diastolic function with less bias then humans show between each other. This system may be useful to allow rapid high-precision bias-resistant measurements of cardiac function.
Supervisor: Francis, Darrel ; Mayet, Jamil ; Hughes, Alun Sponsor: British Heart Foundation
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral