Joao Tourais

Accurate segmentation of cardiac MRI is essential for assessment of cardiac function through biomarkers such as the left and right ventricular ejection fraction (LVEF, RVEF). Although AI methods have achieved high average segmentation accuracy, the precision of biomarkers for individual patients – quantified by estimation variance, remains critical for reliable diagnosis. Calibrated biomarkers, whose uncertainty accurately reflects the true variability, are highly desirable. However, existing evaluations predominantly focus on population-level segmentation accuracy, leaving biomarker-level uncertainty and calibration largely underexplored. Intrinsic anatomical ambiguity and annotation variability are major sources of biomarker variability and cannot be fully eliminated, even when training on a single annotation set. To address this, we propose a probabilistic segmentation framework that explicitly models aleatoric uncertainty with the goal of improving calibration in the biomarker space. The framework disentangles two key sources of uncertainty: (1) detection uncertainty, arising from ambiguous inclusion of basal or apical slices in 2D cardiac MRI, modeled via objectness probabilities; and (2) contour uncertainty, reflecting variability in ventricular boundary delineation, modeled through mean–variance regression of elliptic Fourier descriptors, a compact representation of closed contours. By propagating these uncertainties to derived biomarkers, the proposed method produces more informative and better-calibrated confidence estimates for ejection fraction. Compared to conventional pixel-wise approaches, our framework improves biomarker reliability, particularly in realistic settings dominated by annotation ambiguity and limited domain shift.

Purpose: To optimize Relaxation along a Fictitious Field (RAFF) pulses for rotating frame relaxometry with improved robustness in the presence of and field inhomogeneities.Methods: The resilience of RAFF pulses against and inhomogeneities was studied using Bloch simulations. A parameterized extension of the RAFF formulation was introduced and used to derive a generalized inhomogeneity‐resilient RAFF (girRAFF) pulse. RAFF and girRAFF preparation efficiency, defined as the ratio of the longitudinal magnetization before and after the preparation (), were simulated and validated in phantom experiments. and parametric maps were acquired at 3T in phantom, the calf muscle, and the knee cartilage of healthy subjects. The relaxation time maps were analyzed for resilience against artificially induced field inhomogeneities and assessed in terms of in vivo reproducibility.Results: Optimized girRAFF preparations yielded improved preparation efficiency (0.95/0.91 simulations/phantom) with respect to RAFF (0.36/0.67 simulations/phantom). preparations showed in phantom/calf 6.0/4.8 times higher resilience to inhomogeneities than RAFF, and a 4.7/5.3 improved resilience to inhomogeneities. In the knee cartilage, (53 14 ms) was higher than (42 11 ms). Moreover, girRAFF preparations yielded 7.6/4.9 times improved reproducibility across / inhomogeneity conditions, 1.9 times better reproducibility across subjects and 1.2 times across slices compared with RAFF. Dixon‐based fat suppression led to a further 15‐fold improvement in the robustness of girRAFF to inhomogeneities.Conclusions: RAFF pulses display residual sensitivity to off‐resonance and pronounced sensitivity to inhomogeneities. Optimized girRAFF pulses provide increased robustness and may be an appealing alternative for applications where resilience against field inhomogeneities is required.

Purpose: To develop and evaluate a robust cardiac mapping sequence at 3 T, using Bloch–Siegert shift (BSS)‐based preparations.Methods: A longitudinal magnetization preparation module was designed to encode . After magnetization tip‐down, off‐resonant Fermi pulses, placed symmetrically around two refocusing pulses, induced BSS, followed by tipping back of the magnetization. Bloch simulations were used to optimize refocusing pulse parameters and to assess the mapping sensitivity. Relaxation‐induced error was simulated for various / times. The effective mapping range was determined in phantom experiments, and maps were compared to the conventional BSS method and subadiabatic hyperbolic‐secant 8 (HS8) pulse‐sensitized method. Cardiac maps were acquired in healthy subjects, and evaluated for repeatability and imaging plane intersection consistency. The technique was modified for three‐dimensional (3D) acquisition of the whole heart in a single breath‐hold, and compared to two‐dimensional (2D) acquisition.Results: Simulations indicate that the proposed preparation can be tailored to achieve high mapping sensitivity across various ranges, with maximum sensitivity at the upper range. /‐induced bias did not exceed 5.2. Experimentally reproduced sensitization closely matched simulations for (mean difference 0.0310.022, compared to 0.0180.025 in the HS8‐sensitized method), and showed 20‐fold reduction in the standard deviation of repeated scans, compared with conventional BSS mapping, and an equivalent 2‐fold reduction compared with HS8‐sensitization. Robust cardiac map quality was obtained, with an average test‐retest variability of 0.0270.043 relative to normalized magnitude, and plane intersection bias of 0.0520.031. 3D acquisitions showed good agreement with 2D scans (mean absolute deviation 0.0550.061).Conclusion: BSS‐based preparations enable robust and tailorable 2D/3D cardiac mapping at 3 T in a single breath‐hold.

Deep learning (DL)-based methods have achieved state-of-the-art performance for many medical image segmentation tasks. Nevertheless, recent studies show that deep neural net- works (DNNs) can be miscalibrated and overconfident, leading to ”silent failures” that are risky for clinical applications. Bayesian DL provides an intuitive approach to DL failure de- tection, based on posterior probability estimation. However, the posterior is intractable for large medical image segmentation DNNs. To tackle this challenge, we propose a Bayesian learning framework using Hamiltonian Monte Carlo (HMC), tempered by cold posterior (CP) to accommodate medical data augmentation, named HMC-CP. For HMC compu- tation, we further propose a cyclical annealing strategy, capturing both local and global geometries of the posterior distribution, enabling highly efficient Bayesian DNN training with the same computational budget as training a single DNN. The resulting Bayesian DNN outputs an ensemble segmentation along with the segmentation uncertainty. We evaluate the proposed HMC-CP extensively on cardiac magnetic resonance image (MRI) segmentation, using in-domain steady-state free precession (SSFP) cine images as well as out-of-domain datasets of quantitative T1 and T2 mapping. Our results show that the proposed method improves both segmentation accuracy and uncertainty estimation for in- and out-of-domain data, compared with well-established baseline methods such as Monte Carlo Dropout and Deep Ensembles. Additionally, we establish a conceptual link between HMC and the commonly known stochastic gradient descent (SGD) and provide general insight into the uncertainty of DL. This uncertainty is implicitly encoded in the training dynamics but often overlooked. With reliable uncertainty estimation, our method provides a promising direction toward trustworthy DL in clinical applications. We release our code in https://gitlab.tudelft.nl/yidongzhao/hmc_uncertainty

Objective: Quantitative Magnetic Resonance Imaging (MRI) holds great promise for the early detection of cartilage deterioration. Here, a Magnetic Resonance Fingerprinting (MRF) framework is proposed for comprehensive and rapid quantification of $T_{1}$ , $T_{2}^*$ , and $T_{RAFF2}$ with whole-knee coverage. Methods: A MRF framework was developed to achieve quantification of Relaxation Along a Fictitious Field in the $2^{nd}$ rotating frame of reference ( $T_{RAFF2}$ ) along with $T_{1}$ and $T_{2}^*$ . The proposed sequence acquires 65 measurements of 25 high-resolution slices, interleaved with 7 inversion pulses and 40 RAFF2 trains, for whole-knee quantification in a total acquisition time of 3:25 min. Comparison with reference $T_{1}$ , $T_{2}^*$ , and $T_{RAFF2}$ methods was performed in phantom and in seven healthy subjects at 3T. Repeatability (test-retest) with and without repositioning was also assessed. Results: Phantom measurements resulted in good agreement between MRF and the reference with mean biases of -54, 2, and 5 ms for $T_{1}$ , $T_{2}^*$ , and $T_{RAFF2}$ , respectively. Complete characterization of the whole-knee cartilage was achieved for all subjects, and, for the femoral and tibial compartments, a good agreement between MRF and reference measurements was obtained. Across all subjects, the proposed MRF method yielded acceptable repeatability without repositioning ( $R^{2}\geq$ 0.94) and with repositioning ( $R^{2}\geq$ 0.57) for $T_{1}$ , $T_{2}^*$ , and $T_{RAFF2}$ . Significance: The short scan time combined with the whole-knee coverage makes the proposed MRF framework a promising candidate for the early assessment of cartilage degeneration with quantitative MRI, but further research may be warranted to improve repeatability after repositioning and assess clinical value in patients.

Purpose: To develop a free‐breathing (FB) 2D radial balanced steady‐state free precession cine cardiac MRI method with 100% respiratory gating efficiency using respiratory auto‐calibrated motion correction (RAMCO) based on a motion‐sensing camera.Methods: The signal from a respiratory motion‐sensing camera was recorded during a FB retrospectively electrocardiogram triggered 2D radial balanced steady‐state free precession acquisition using pseudo–tiny‐golden‐angle ordering. With RAMCO, for each acquisition the respiratory signal was retrospectively auto‐calibrated by applying different linear translations, using the resulting in‐plane image sharpness as a criterium. The auto‐calibration determines the optimal magnitude of the linear translations for each of the in‐plane directions to minimize motion blurring caused by bulk respiratory motion. Additionally, motion‐weighted density compensation was applied during radial gridding to minimize through‐plane and non‐bulk motion blurring. Left ventricular functional parameters and sharpness scores of FB radial cine were compared with and without RAMCO, and additionally with conventional breath‐hold Cartesian cine on 9 volunteers.Results: FB radial cine with RAMCO had similar sharpness scores as conventional breath‐hold Cartesian cine and the left ventricular functional parameters agreed. For FB radial cine, RAMCO reduced respiratory motion artifacts with a statistically significant difference in sharpness scores (P < 0.05) compared to reconstructions without motion correction.Conclusion: 2D radial cine imaging with RAMCO allows evaluation of left ventricular functional parameters in FB with 100% respiratory efficiency. It eliminates the need for breath‐holds, which is especially valuable for patients with no or impaired breath‐holding capacity. Validation of the proposed method on patients is warranted.

Background: Respiratory gating is generally recommended in 4D flow MRI of the heart to avoid blurring and motion artifacts. Recently, a novel automated contact‐less camera‐based respiratory motion sensor has been introduced.Purpose: To compare camera‐based respiratory gating (CAM) with liver‐lung‐navigator‐based gating (NAV) and no gating (NO) for whole‐heart 4D flow MRI.Study Type: Retrospective.Subjects: Thirty two patients with a spectrum of cardiovascular diseases.Field Strength/Sequence: A 3T, 3D‐cine spoiled‐gradient‐echo‐T1‐weighted‐sequence with flow‐encoding in three spatial directions.Assessment: Respiratory phases were derived and compared against each other by cross‐correlation. Three radiologists/cardiologist scored images reconstructed with camera‐based, navigator‐based, and no respiratory gating with a 4‐point Likert scale (qualitative analysis). Quantitative image quality analysis, in form of signal‐to‐noise ratio (SNR) and liver‐lung‐edge (LLE) for sharpness and quantitative flow analysis of the valves were performed semi‐automatically.Statistical Tests: One‐way repeated measured analysis of variance (ANOVA) with Wilks's lambda testing and follow‐up pairwise comparisons. Significance level of P ≤ 0.05. Krippendorff's‐alpha‐test for inter‐rater reliability.Results: The respiratory signal analysis revealed that CAM and NAV phases were highly correlated (C = 0.93 ± 0.09, P < 0.01). Image scoring showed poor inter‐rater reliability and no significant differences were observed (P ≥ 0.16). The image quality comparison showed that NAV and CAM were superior to NO with higher SNR (P = 0.02) and smaller LLE (P < 0.01). The quantitative flow analysis showed significant differences between the three respiratory‐gated reconstructions in the tricuspid and pulmonary valves (P ≤ 0.05), but not in the mitral and aortic valves (P > 0.05). Pairwise comparisons showed that reconstructions without respiratory gating were different in flow measurements to either CAM or NAV or both, but no differences were found between CAM and NAV reconstructions.Data Conclusion: Camera‐based respiratory gating performed as well as conventional liver‐lung‐navigator‐based respiratory gating. Quantitative image quality analysis showed that both techniques were equivalent and superior to no‐gating‐reconstructions. Quantitative flow analysis revealed local flow differences (tricuspid/pulmonary valves) in images of no‐gating‐reconstructions, but no differences were found between images reconstructed with camera‐based and navigator‐based respiratory gating.Level of Evidence: 3Technical Efficacy: Stage 2

Purpose: To develop a new 3D radial trajectory based on the natural spiral phyllotaxis (SP), with variable anisotropic FOV.Theory & Methods: A 3D radial trajectory based on the SP with favorable interleaving properties for cardiac imaging has been proposed by Piccini et al (Magn Reson Med. 2011;66:1049‐1056), which supports a FOV with a fixed anisotropy. However, a fixed anisotropy can be inefficient when sampling objects with different anisotropic dimensions. We extend Larson’s 3D radial method to provide variable anisotropic FOV for spiral phyllotaxis (VASP). Simulations were performed to measure distance between successive projections, analyze point spread functions, and compare aliasing artifacts for both VASP and conventional SP. VASP was fully implemented on a whole‐body clinical MR scanner. Phantom and in vivo cardiac images were acquired at 1.5 tesla.Results: Simulations, phantom, and in vivo experiments confirmed that VASP can achieve variable anisotropic FOV while maintaining the favorable interleaving properties of SP. For an anisotropic FOV with 100:100:35 ratio, VASP required ~65% fewer radial projections than the conventional SP to satisfy Nyquist criteria. Alternatively, when the same number of radial projections were used as in conventional SP, VASP produced fewer aliasing artifacts for anisotropic objects within the excited imaging volumes.Conclusion: We have developed a new method (VASP), which enables variable anisotropic FOV for 3D radial trajectory with SP. For anisotropic objects within the excited imaging volumes, VASP can reduce scan times and/or reduce aliasing artifacts.