BACKGROUND Cardiac magnetic resonance (CMR) sequences necessitate synchronisation with the cardiac cycle. This is routinely obtained through electrocardiographic (ECG) detection of the R-wave with subsequent triggering or gating of the acquisition, depending on the type of sequence. An accepted alternative is the use of transmissive photoplethysmography using a fingerclip (fPPG) to detect the peripheral pulse wave. OBJECTIVES To assess the feasibility of contact-free remote photoplethysmography (rPPG) for CMR triggering/gating. METHODS In this prospective study (www.drks.de, DRKS00036686), CMR sequences were triggered/gated by ECG, fPPG and rPPG. The primary endpoint was trigger accuracy, measured by true positive rate (TPR), positive predictive value (PPV), beat detection error rate (BDER) and trigger-latency variability (TLV) in the rPPG and fPPG tracings compared to the ECG tracing. Secondary endpoints were technical success, image quality and quantitative CMR measurements of the acquired CMR sequences. Subsequently, the rPPG algorithm was adjusted to improve pulse wave detection in patients with skin type Fitzpatrick 5-6. RESULTS 64 subjects were included. rPPG initially provided comparable measures of trigger accuracy to fPPG, with worse performance in subjects with skin type Fitzpatrick 5-6. Adjusting the rPPG algorithm significantly improved trigger accuracy irrespective of skin type: BDER (%) for fPPG 8.5 (0.9-23.0), for rPPG 7.2 (0.3-49.1) and for the adjusted rPPG 1.3 (0.2-11.2). Image quality with both fPPG and rPPG were comparable to ECG. The differences observed in quantitative CMR measurements between fPPG, rPPG and ECG were in the order of inter-observer-variability. CONCLUSIONS Contact-free rPPG is a viable alternative to standard CMR triggering/gating using ECG or fPPG. Adjustment of our algorithm led to excellent trigger accuracy irrespective of skin type.
ABSTRACT: Background: Four-dimensional (4D) flow magnetic resonance imaging (MRI) hemodynamic biomarkers (e.g., peak velocity and wall shear stress [WSS]) have shown promise for improved risk-stratification in patients with thoracic aortic disease (TAD). However, lengthy 4D flow scan times, complex data analysis and need for dedicated expertise limit clinical translation. In this study, we developed a fluid-physics informed deep generative adversarial neural network, CE-MRA-FLOWnet, to predict aortic hemodynamics directly from standard-of-care contrast-enhanced magnetic resonance angiography (CE-MRA) images. Methods: We retrospectively identified 1392 patients (age: 52±12years, 1011 male, 954 with bicuspid aortic valve, BAV; 438 with tricuspid aortic valve, TAV) who underwent paired clinical 4D flow MRI and CE-MRA between 2011 and 2020 for suspected TAD. The CE-MRA-FLOWnet used CE-MRA data (1127 for network training, 265 for testing) to generate a prediction of aortic hemodynamics. 4D flow-measured aortic systolic three-dimensional (3D) blood flow velocity vector fields served as ground truth. Analysis included comparison of AI-derived and ground truth aortic systolic peak velocity (PV) and WSS, the relative area of the ascending aorta (AAo, %) exposed to elevated WSS, and aortic valve stenosis severity grading. The predictive value of CE-MRA-FLOWnet hemodynamics metrics was assessed in a subgroup of 133 BAV patients with known multi-year adverse outcomes. Results: CE-MRA-FLOWnet training time was 8100 min; inference time per CE-MRA was 0.88±0.05 seconds. AI-derived PV showed strong regional agreement and low biases with ground truth 4D flow (0.00–0.04 m/s), and relative differences within 10.6%-12.7%. Aortic WSS also demonstrated minimal bias (-0.01–0.04 Pa) and close alignment of the area of elevated WSS between AI and 4D flow (19.6±15.1% vs. 19.3±16.2%, p=0.84). AS severity was accurately classified in 88% (233/265) of cases, with all grading errors limited to a one-grade difference (Kappa 0.84). Receiver operating characteristic (ROC) analysis showed that CE-MRA-FLOWnet derived hemodynamic metrics outperformed diameter alone for predicting adverse outcomes (PV area under the curve [AUC] = 0.73–0.76; WSS = 0.83–0.88; diameter = 0.52–0.62). Conclusion: CE-MRA-FLOWnet accurately predicts aortic hemodynamics in TAD patients using standard-of-care CE-MRA images. Our results demonstrate potential for clinical integration by providing physicians with near real-time hemodynamic data from widely available clinical CE-MRA images.
