Diagnostic and prognostic value of coronary artery flow and morphology in a multicentre randomised trial of computed tomography versus invasive angiography: clinical radiomics analysis

• Applicant: Professor Dr. Marc Dewey, since 2/2023
• Subject Area: Nuclear Medicine, Radiotherapy, Radiobiology ; Medical Physics, Biomedical Technology
• Term: from 2019 to 2024
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 428223139

Background and Objective: Computed tomography angiography (CTA) allows reliably and noninvasively excluding coronary artery stenosis when using invasive coronary angiography (ICA) as the diagnostic reference standard. There is also early evidence that coronary flow values can be derived from CTA using computational fluid dynamics (CFD) or machine learning (ML) with accuracy similar to adenosine-based flow reserve in ICA. By applying similar models of CFD to the morphological information from ICA, the quantitative flow ratio (QFR) can be computed without the necessity of a pressure wire. However, the prognostic value of CTA using CFD or ML and ICA using QFR has not been determined in a randomised trial; yet such a randomised comparison would be crucial for clinical translation and implementation. We propose to analyze CTA using advanced image processing with prospectively obtained data from randomised trials. In this project, we will explore the capabilities of CFD and ML to improve diagnosis and prognosis of coronary CTA in patients with suspected coronary artery stenosis. Methods and Work Programme: We conducted two randomised trials of CTA and ICA in patients with suspected coronary artery stenosis: the single-centre CAD-Man study in 340 patients and the multicentre DISCHARGE trial in 3546 patients. Both trials included an initial diagnostic management phase after CTA and ICA as well as a long-term clinical follow-up after a maximum of four years which provides clearly defined endpoints for comparing the diagnostic and prognostic value of CT using CFD and ML and ICA using QFR. To achieve this comparison we will first prepare and pre-process existing imaging data from patients randomised to CTA and ICA in DISCHARGE. This ensures homogenisation of CTA and ICA data, which stem from 26 different European sites, to achieve robust and generalizable results for stenosis and flow reserve quantification. Second, we will conduct feature extraction for advanced image processing of CTA and ICA data from the DISCHARGE trial data for stenosis and flow quantification. Third, these features will be used in the core of this project, which is the implementation and exploration of the radiomics analysis to compare the diagnostic and prognostic value of CTA and ICA. This will be done using endpoints prospectively defined including stenosis and flow results for diagnostic values as well as myocardial infarction, stroke, and cardiovascular death for prognostic value. The quantitative results of image analysis will be made available as a database within and beyond the SPP. Anticipated Gain of Knowledge: We anticipated novel insights into the association of CTA-derived features of coronary flow and morphology with diagnostic and prognostic outcomes of patients. The randomised comparison with ICA ensures that the results can lead to direct clinical translation and implementation of coronary CTA in the work-up for patients with suspected coronary artery stenosis.

• DFG Programme: Priority Programmes
• Subproject of: SPP 2177: Radiomics: Next Generation of Biomedical Imaging
• Ehemalige Antragstellerin: Privatdozentin Dr. Elke Zimmermann, until 1/2023