Lo, Ernest; (2021) Integration of patient-specific myocardial perfusion in CT-based FFR computations. Doctoral thesis (Ph.D), UCL (University College London).

Preview

Text ELoThesis21.pdf Download (4MB) | Preview

Abstract

Computed Tomography based Fractional Flow Reserve (FFRCT) is a non-invasive simulation based measure for diagnosing ischaemia causing arterial stenoses. One drawback of simulation based measures are the assumptions made that are usually based on population studies that may not apply to all patients. This study describes the fundamental characteristics to FFRCT simulations and how the simulations can be simplified where it can and where assumptions break down. The investigation starts with assessing whether the simulations can be simplified to a steady flow, whilst uncharacteristic of typical coronary blood flow, it was demonstrated that with regards to the diagnostic measures of FFR, and its variants dFFR or iFR, that steady flow was applicable, which reduces the complexity of the simulation, saving computational time and resources as well as removing uncertainty in the input assumptions.[1] The next phase of the study explored the downstream conditions of the FFRCT simulation scheme. The microvasculature is too small to resolve in CT imaging and therefore assumptions are made regarding its form and function. Whilst form function relationships of the microvasculature are well established in the literature for the structure of microvessels at rest, assumptions regarding stress or hyperaemia are used for FFRCT to simulate maximal blood flow through the coronary arteries. The investigation utilised perfusion imaging to assess the validity of this assumption and showed how variable the microvascular response to hyperaemia is, and the effect that has on FFRCT.[2][3] The last part of the study produced a novel method of estimating the microvascular response using patient metrics such as age, sex, diabetes, smoker status etc, from a training dataset of 101 patients. By using the patient-specific microvascular response, FFRCT simulations better represent the coronary artery health of the patient. On a separate dataset of 10 patients, the FFRCT measurements using this novel method was also validated against the gold standard invasive FFR and has demonstrated a better diagnostic performance (94% accuracy) than the conventional method (82% accuracy). Secondly the novel method also created a probabilistic spread of FFRCT values that may provide better utility than a strict binary measure. Whilst this novel method will require further validation with larger studies, it nevertheless has potential to address some of the current drawbacks of FFRCT methods when applied to a varied patient demographic.

Download activity - last month

Export as JSONXMLCSV

Download activity - last 12 months

Export as CSVXMLJSON

Downloads by country - last 12 months

Export as CSVXMLJSON

Archive Staff Only

View Item