Analysis identifying minimal governing parameters for clinically accurate in silico fractional flow reserve

Abstract

BackgroundPersonalized hemodynamic models can accurately compute fractional flow reserve (FFR) from coronary angiograms and clinical measurements (FFRbaseline), but obtaining patient-specific data could be challenging and sometimes not feasible. Understanding which measurements need to be patient-tuned vs. patient-generalized would inform models with minimal inputs that could expedite data collection and simulation pipelines.AimsTo determine the minimum set of patient-specific inputs to compute FFR using invasive measurement of FFR (FFRinvasive) as gold standard.Materials and MethodsPersonalized coronary geometries (N=50) were derived from patient coronary angiograms. A computational fluid dynamics framework, FFRbaseline, was parameterized with patient-specific inputs: coronary geometry, stenosis geometry, mean arterial pressure, cardiac output, heart rate, hematocrit, and distal pressure location. FFRbaseline was validated against FFRinvasive and used as the baseline to elucidate the impact of uncertainty on personalized inputs through global uncertainty analysis. FFRstreamlined was created by only incorporating the most sensitive inputs and FFRsemi-streamlined additionally included patient-specific distal location.ResultsFFRbaseline was validated against FFRinvasive via correlation (r=0.714, p<0.001), agreement (mean difference: 0.01±0.09), and diagnostic performance (sensitivity: 89.5%, specificity: 93.6%, PPV: 89.5%, NPV: 93.6%, AUC: 0.95). FFRsemi-streamlined provided identical diagnostic performance with FFRbaseline. Compared to FFRbaseline vs. FFRinvasive, FFRstreamlined vs. FFRinvasive had decreased correlation (r=0.64, p<0.001), improved agreement (mean difference: 0.01±0.08), and comparable diagnostic performance (sensitivity: 79.0%, specificity: 90.3%, PPV: 83.3%, NPV: 87.5%, AUC: 0.90).ConclusionStreamlined models could match the diagnostic performance of the baseline with a full gamut of patient-specific measurements. Capturing coronary hemodynamics depended most on accurate geometry reconstruction and cardiac output measurement.

Enhanced CT simulation using realistic vascular flow dynamics

“…We used our high-resolu�on 1D blood flow simulator, HARVEY1D [5], [6], to model flow. As 1D models only resolve flow in the longitudinal direc�on of flow, contrast is assumed to be constant over cross-sec�ons.…”

“…At each �mestep, once the full flow field is computed, the coupled flow model solves 1D transport equa�ons [7] that advect volume frac�ons of contrast. To accurately model the effects of stenoses, we coupled the 1D model to an empirically-derived pressure-loss equa�on that has been validated for coronaries [6], [8].…”

The Potential of Deep Learning to Advance Clinical Applications of Computational Biomechanics

“…Coronary-artery fractional -flow reserve [40] Identification of minimal number of patient variables to estimate fractional flow reserve While these models can lead to improved diagnostic criteria, a number of limitations need to be addressed. These include the limitations of the imaging technology in the provision of sufficient resolution, the need to validate the model with clinical data, detailed information about loading conditions, and the need for models to incorporate features that depend on age, sex, and race in order to account for population variability [41].…”