The computational efficiency of contemporary multi-scale digital twins used to investigate cardiac arrhythmia is often hindered by the equation system complexity, while the accuracy of predictions depends on the calibration of model parameters and their intricate non-linear relation. Here we employed a robust pipeline for fast simulations of scar-related ventricular tachycardia (VT) inducibility using an efficient Lattice-Boltzmann Method with GPU-based accelerated code. We tested the pipeline on 3D digital twins built from MR images acquired in n = 8 swine with chronic infarction, by performing a subject-specific personalization of each model per tissue type (i.e., scar, border zone, and healthy) and tuning the key parameters (e.g., action potential duration, excitability, and wave speed) from recorded endocardial bipolar voltage maps and intracardiac electrograms. We first validated the VT simulation outcome by precisely replicating the stimulation protocol and pacing site used in the animal studies. Our results demonstrated very good agreement between the experiment and simulated VT outcome for personalized model parameters using subject-specific values, compared to a poor outcome when parameters were tuned from average values from all cases. Second, we performed a comprehensive in silico study where the stimulation was delivered from 10 000 virtual endocardial locations and found a strong dependence of VT (non)/-inducibility per case on the stimulation site. Simulating 10 s of sustained VT induced on a 3D model comprising 1.2 million cubic voxels (0.7 mm edge size) took 30 min on a laptop using GPU, underlying the computational tractability of our pipeline and its potential clinical translation to predict infarct-related VT risk.

Endovascular treatment of complex peripheral artery disease and chronic total occlusions remains a significant challenge. This is largely due to limited distal control of devices used and reliance on proximal manipulation with conventional catheters. To bridge this gap, we previously developed the CathPilot, a modular steerable catheter that integrates a self-expanding nitinol frame for local anchoring with a real-time graphical user interface that provides cross-sectional tip position feedback alongside fluoroscopy. This study compares the CathPilot with conventional non-steerable catheters in a simulated-use, within-subjects design. Eight interventional clinicians each performed three lesion-crossing attempts per device in a silicone phantom model of human vasculature. Silicone lesions were positioned in the left superficial femoral artery. Objective endpoints included lesion-crossing success, procedure time, fluoroscopy time, and radiation dose. Subjective assessments included the System Usability Scale (SUS) for perceived usability and NASA Task Load Index (NASA-TLX) for operator workload. Combining objective performance with human-factors instruments enabled a comprehensive assessment of the CathPilot’s procedural usability against conventional catheters. CathPilot achieved a 92

Background Late gadolinium enhancement magnetic resonance imaging (MRI) has been shown to reliably locate radiofrequency ablation (RFA) lesions with microvascular obstruction (MVO) as a broadly accepted signature. However, MVO volume depends on the time elapsed after gadolinium contrast injection. Native T1-weighted (T1w) MRI has recently been shown to accurately characterize RFA lesions in preclinical models. Objective This study aimed to demonstrate the feasibility of wideband native T1w MRI in characterizing lesion dimensions in patients within 7 days of ventricular tachycardia RFA. Methods Postablation wideband native T1w and 3-dimensional late gadolinium enhancement MRI was performed within 7 days of RFA. RFA lesion surface area (mm2), volume (mm3), and native T1w maximum lesion depth were calculated. Average catheter contact force (g), ablation duration (seconds), and impedance drop (Ω) were calculated. Patients with and without implantable cardioverter-defibrillators were included. Results Ten patients (median age = 64.5 years, 60% ischemic) underwent ventricular tachycardia RFA (80% endocardial, 20% epicardial RFA) and postablation cardiac MRI within 7 days of RFA (median = 4.5 days). There was a significantly greater mean MVO surface area and volume than mean native T1w lesion surface area and volume (P = .0108, P = .0376, P < .05), respectively. There were strong positive correlations between average surface area and average volume measured by MVO and native T1w (r = 0.71, P = .0237, r = 0.70, P = .0265, P < .05), respectively. Average maximum native T1w lesion depth correlated with average ablation duration and average impedance drop (r = 0.73, P = .02, r = 0.68, P = .04, P < .05), respectively. Conclusion Native T1w MRI can effectively characterize lesion dimensions in patients with and without implantable cardioverter-defibrillators within 7 days after ablation. This emerging biomarker may facilitate early prediction of RFA efficacy and ablation success particularly in patients demonstrating contraindications to gadolinium contrast agents.

Objective: Endovascular revascularization of peripheral arterial occlusions has a high technical failure rate of 15-20%, mainly due to difficulties in crossing the occlusion with a guidewire. This study evaluates the use of a Picosecond mid-Infrared Laser (PIRL) to facilitate occlusion crossing. Methods: Popliteal artery lesion samples were obtained from a donated limb of a patient with critical limb ischemia (CLI). A customized system advanced the PIRL fiber at controlled speeds toward the occlusion. The fiber was tested with its source OFF and ON at either 500 mW or 1000 mW power, 2.96 mu m wavelength, and 1 kHz repetition rate. Lesions were scanned using mu -CT before and after the test, and post-ablated tissues were analyzed histologically. The feasibility of using PIRL with the CathCam, an optical image-guided steerable catheter, was also assessed under X-ray fluoroscopy in an OR suite. Results: Tests showed a significant crossing success improvement with the laser ON vs. OFF (95.6% vs. 73.9%, p << 0.05) and a significant reduction in maximum force (5.5 +/- 9.8 gr vs. 17.2 +/- 12.3 gr; p << 0.05). Success rates generally decreased with increased fiber speed, ranging from 100% at 0.019 mm/s to 30% at 0.5 mm/s, while force increased. The results showed that 0.1 mm/s fiber advancement speed is the fastest speed with the highest crossing success rate. Histological analysis showed sub -50 mu m tissue trauma post-PIRL-ablation. Conclusion: PIRL plaque ablation is minimally invasive, and 0.1 mm/s was identified as the optimal fiber advancement speed. Significance: PIRL, guided with CathCam, demonstrates high potential for endovascular revascularization procedures.

PURPOSE:True real-time cardiac MRI (CMR), necessary for capturing live cardiac dynamics and imaging irregular cardiac rhythms, remains challenging. In this article, we move toward real-time CMR in multiple reconstruction frameworks via strategies to predict cardiac motion, improve computational efficiency, reduce artifacts, and preserve spatial resolution. THEORY AND METHODS:A published predictive signal model (PMOT) for imaging irregular cardiac dynamics was modified (mPMOT) to enable efficient computation of state-transition matrices for predicting cardiac motion, as training PMOT is computationally expensive. A multi-rate Kalman filter framework was developed to enable computationally efficient reconstructions of high-resolution, large-matrix CMR datasets. Reconstructions were evaluated on multi-coil CMR data in human and swine using multi-rate Kalman filtering and compressed sensing (CS). RESULTS:Training mPMOT is two orders of magnitude faster than PMOT. Across all datasets and frameworks, mPMOT facilitated high-quality reconstructions of CMR images for different undersampling patterns at acceleration factors of 9 and 13.5. Furthermore, mPMOT substantially reduced temporal blurring artifacts naturally present in CS reconstructions. In swine, mPMOT reduced the mean-squared error of the multi-rate Kalman filter by two orders of magnitude. The multi-rate Kalman filter implementation maintained spatial resolution while reducing computation time from 5439 s to 56 s in select applications. CONCLUSION:Our mPMOT is computationally efficient and can be integrated within multiple established reconstruction frameworks to ensure robust tracking and reconstruction for dynamic and real-time CMR applications.

Background The role of advanced (cardiac magnetic resonance [CMR] or positron emission tomography [PET]) vs single-photon emission computerized tomography (SPECT) ischemia imaging to guide management remains unclear in patients with ischemic heart failure (IHF). The primary aim was to determine the effect of imaging modality on a composite cardiovascular endpoint and cardiac death in patients with IHF who require ischemia assessment. Methods Patients with IHF were randomized to advanced or SPECT imaging. A parallel registry also was performed. The primary endpoint was the composite of cardiac death, infarction, arrest, and cardiac rehospitalization. The key secondary endpoint was cardiac death. Results Patients in the randomized population (advanced imaging [PET or CMR; n = 64] or SPECT [n = 56]) had a cumulative incidence rate (CIR) for the primary endpoint of 33.1% and 33.0%, respectively (hazard ratio [HR] 0.94, 95% confidence interval [CI] 0.49, 1.80, P = 0.853). CIRs for cardiac death were 13.8% and 25.1%, respectively (HR 0.62, 95% CI 0.25, 1.80, P = 0.296).In the parallel registry (n = 336 advanced; n = 216 SPECT), the primary endpoint CIRs were 31.2% and 35.3%, respectively (HR 0.81, 95% CI 0.56, 1.19, P = 0.284). CIRs for cardiac death were 11.0% and 16.6%, respectively (HR 0.53, 95% CI 0.27, 1.04, P = 0.066). Patients were followed for a median (interquartile range) of 24.1 (11.6, 27.5) months.Pooled analysis from the randomized and registry populations revealed a significant benefit of advanced imaging for reduction of cardiac death (HR 0.56, 95% CI 0.33, 0.96, P = 0.04) with minimal heterogeneity (I2 = 0%). Conclusion Among IHF patients assessed for ischemia, advanced imaging (PET or CMR) was not associated with reduced composite cardiac events, compared to SPECT. Clinical Trial Registration NCT01288560.

PURPOSE:Post-contrast T 1 ∗ $$ {T}_1^{\ast } $$ mapping has proven promising for automated scar segmentation in subjects without ICDs, but this has not been implemented in patients with ICDs. We introduce an automated cluster-based thresholding method for T 1 ∗ $$ {T}_1^{\ast } $$ maps with an ICD present and compare it to manually tuned thresholding of synthetic LGE images with an ICD present and standard LGE without an ICD present. METHODS:Seven swine received an ischemia-reperfusion myocardial infarction and were imaged at 3 T 4-5 weeks post-infarct with and without an ICD. Mapping-based thresholding was performed using synthetic LGE and artifact-corrected cluster-thresholding methods, both employing connected component filtering. Standard pixel signal intensity thresholding was performed on the conventional LGE without an ICD. Volumetric accuracy is relative to conventional LGE and Dice similarity between SynLGE and cluster-based segmentations were evaluated. RESULTS:No statistical significance was observed between LGE volumes without an ICD and both SynLGE and artifact-corrected cluster-threshold volumes with an ICD, when using connected component filtering. Additionally, Dice alignment between SynLGE and cluster-thresholding was high for healthy myocardium (0.96), dense scar (0.83), and dense scar union gray zone (0.91) when artifact correction and connected component filtering were implemented. CONCLUSION:Clustering of T 1 ∗ $$ {T}_1^{\ast } $$ maps holds promise for a reproducible approach to scar segmentation in the presence of ICDs.

Background: Sudden cardiac death is a leading worldwide cause of cardiac mortality and is largely related to ventricular tachycardia (VT) in patients with known myocardial scarring. In these patients, implantable cardioverter-defibrillator (ICD) therapy reduces arrhythmia-related mortality. However, curative procedures such as catheter ablation are used to homogenize regions of scar and remove structural re-entry circuits that cause VT. Objective: In this paper, we conduct a preliminary experiment comparing 2-dimensional (2D) late gadolinium enhancement (LGE) and 3-dimensional (3D) LGE without an ICD with wideband motion-corrected (WB-MOCO) LGE with an ICD in a cohort of infarcted Yorkshire swine. Methods: Animals were imaged after infarct with conventional 2D and 3D LGE without an ICD present and 2D WB-MOCO LGE with an ICD present. Images were analyzed to determine heterogeneous tissue corridor (HTC) count and location, which were compared with circuit exit locations determined using a 12-lead electrocardiogram. Results: We found a statistically significant increase in HTC count with WB-MOCO LGE, but no significant differences in the number of true-positive or false-positive HTCs per subject. Conclusion: WB-MOCO LGE has reduced specificity to physiologically relevant HTCs than conventional 2D or 3D LGE.

Cardiovascular diseases (CVDs) remain the leading cause of mortality and morbidity worldwide. Both diagnosis and prognosis of these diseases benefit from high-quality imaging, which cardiac magnetic resonance imaging provides. CMR imaging requires lengthy acquisition times and multiple breath-holds for a complete exam, which can lead to patient discomfort and frequently results in image artifacts. In this work, we present a Low-rank tensor U-Net method (LowRank-CGNet) that rapidly reconstructs highly undersampled data with a variety of anatomy, contrast, and undersampling artifacts. The model uses conjugate gradient data consistency to solve for the spatial and temporal bases and employs a U-Net to further regularize the basis vectors. Currently, model performance is superior to a standard U-Net, but inferior to conventional compressed sensing methods. In the future, we aim to further improve model performance by increasing the U-Net size, extending the training duration, and dynamically updating the tensor rank for different anatomies.

BACKGROUND:Various pulsed field ablation (PFA) parameters have been proposed to improve lesion depth. This study evaluated a modified unipolar return PFA system to create deep lesions in healthy and infarcted ventricular myocardia. METHODS:Numerical modeling was used to compare a modified unipolar return PFA system configuration with a conventional unipolar return (skin patch). We then performed ablation in 14 swine (5 with chronic myocardial infarction and 9 healthy). PFA lesions were created in the left ventricle using a focal catheter (4-mm tip) with a return electrode positioned in the inferior vena cava (biphasic, microsecond pulses of 1300 and 1500 V, 1-16 trains). Electroanatomical mapping guided ablation and lesion localization on magnetic resonance imaging were performed 48 hours post-ablation in the infarcted group and at 1 day, 7 days, and 6 weeks post-ablation in the healthy group. RESULTS:Numerical modeling demonstrated that the modified unipolar return PFA system produced deeper lesions with reduced variability compared with the skin patch. In healthy pigs (n=35 lesions), depths of 6.8±1.8 mm and widths of 11.5±4.7 mm were achieved with 8 pulse trains. Depths of 8.2±2.8 mm and widths of 14.0±4.7 mm were achieved with 16 trains. The maximum lesion depths were 8.8 and 11.6 mm for 8 and 16 trains, respectively. In the infarcted cohort (n=22 lesions), all lesions applied to scar tissue penetrated through fibrotic regions, with epicardial involvement observed in 57% of lesions. CONCLUSIONS:The modified unipolar return PFA system effectively creates large lesions and can achieve transmurality in healthy and infarcted animals. Compared with conventional unipolar, it may offer greater lesion depth, width, and consistency.

During endovascular revascularization interventions for peripheral arterial disease, the standard modality of X-ray fluoroscopy (XRF) used for image guidance is limited in visualizing distal segments of infrapopliteal vessels. To enhance visualization of arteries, an image registration technique was developed to align pre-acquired computed tomography (CT) angiography images and to create fusion images highlighting arteries of interest. X-ray image metadata capturing the position of the X-ray gantry initializes a multiscale iterative optimization process, which uses a local-variance masked normalized cross-correlation loss to rigidly align a digitally reconstructed radiograph (DRR) of the CT dataset with the target X-ray, using the edges of the fibula and tibia as the basis for alignment. A precomputed library of DRRs is used to improve run-time, and the six-degree-of-freedom optimization problem of rigid registration is divided into three smaller sub-problems to improve convergence. The method was tested on a dataset of paired cone-beam CT (CBCT) and XRF images of ex vivo limbs, and registration accuracy at the midline of the artery was evaluated. On a dataset of CBCTs from 4 different limbs and a total of 17 XRF images, successful registration was achieved in 13 cases, with the remainder suffering from input image quality issues. The method produced average misalignments of less than 1 mm in horizontal projection distance along the artery midline, with an average run-time of 16 s. The sub-mm spatial accuracy of artery overlays is sufficient for the clinical use case of identifying guidewire deviations from the path of the artery, for early detection of guidewire-induced perforations. The semiautomatic workflow and average run-time of the algorithm make it feasible for integration into clinical workflows.

BACKGROUND:Implantable cardioverter-defibrillators (ICDs) cause banding artifacts around areas of B0 inhomogeneity in conventional steady-state free precession (SSFP) cine sequences. Alternatively, high-bandwidth gradient-recalled echo (GRE) cine sequences can be used to minimize artifacts in the myocardium. In this study, we assessed the bias and interobserver variability in cardiac volumes and ejection fractions between GRE cines in acquired in the presence of ICDS and ground-truth SSFP cines (without ICDs present) in a population of healthy volunteers. Further, a small cohort of ICD patients was recruited and scanned to demonstrate clinical feasibility. METHODS:High-bandwidth GRE cine was performed in 11 healthy volunteers with taped ICDs mimicking clinical implants. After the ICD was removed, ground-truth SSFP cine was performed. Two observers separately assessed image quality metrics and contoured the cine images to return cardiac volumes and ejection fractions. Nine patients with an ICD were also scanned with the GRE cine protocol before contrast administration; data were contoured by two observers and analyzed for interobserver agreement. RESULTS:In the healthy volunteer dataset, no statistically significant differences were found when comparing volumes or ejection fractions between sequences (p > 0.05). Statistically significant differences were found when comparing right ventricular ejection fraction (RVEF) (p = 0.009) and right ventricular end-systolic volume (p = 0.029) between observers, with no other significant interobserver differences. The interobserver variability of patient left ventricular ejection fraction and RVEF data was 3-4%, with lower image quality metrics for patient scans than volunteer scans. CONCLUSION:GRE cine imaging in healthy volunteers with taped ICDs demonstrated good agreement with SSFP cine, but increased interobserver variability. In patients, reducing the breath-hold duration caused a decrease in image quality, with GRE cine imaging in patients with ICDs demonstrating poorer image quality and greater interobserver variability than in healthy volunteer studies. Future work is needed to improve GRE cine image quality in patients with ICDs to reduce interobserver variability and improve clinical confidence.

Personalized computer modeling is a powerful, effective and non-invasive tool, that can be used to simulate the inducibility of ventricular tachycardia (VT) in scar-related cases. In this in silico study we propose a robust pipeline for VT risk prediction based on preclinical digital twins developed and parameterized from high resolution in vivo MR images and electro-anatomical datasets acquired in 5 pigs with chronic infarction. Specifically, we employed a modified Mitchell-Schaeffer computational model and calibrated per case the key parameter that tunes the action potential duration using the recorded intracardiac ECGs. To validate the predictions, we compared the measured VT cycle length (CL) per case to the simulated VTCL obtained from each calibrated digital twin by precisely replicating the experimental inducibility protocol. We further defined 100 pacing sites on the right and left endocardial surfaces, respectively, and investigated the impact of pacing location on the VT inducibility. Overall, results demonstrated that our pipeline was able to accurately predict VT inducibility after calibration, reproducing the experimental VT with a small error in CLs (i.e., <4

PURPOSE:To investigate the feasibility of a deep learning algorithm combining variational autoencoder (VAE) and two-dimensional (2D) convolutional neural networks (CNN) for automatically quantifying hard tissue presence and morphology in multi-contrast magnetic resonance (MR) images of peripheral arterial disease (PAD) occlusive lesions. METHODS:Multi-contrast MR images (T2-weighted and ultrashort echo time) were acquired from lesions harvested from six amputated legs with high isotropic spatial resolution (0.078 mm and 0.156 mm, respectively) at 9.4 T. A total of 4014 pseudo-color combined images were generated, with 75% used to train a VAE employing custom 2D CNN layers. A Gaussian mixture model (GMM) was employed to classify the latent space data into four tissue classes: I) concentric calcified (c), II) eccentric calcified (e), III) occluded with hard tissue (h) and IV) occluded with soft tissue (s). Test image probabilities, encoded by the trained VAE were used to evaluate model performance. RESULTS:GMM component classification probabilities ranged from 0.92 to 0.97 for class (c), 1.00 for class (e), 0.82-0.95 for class (h) and 0.56-0.93 for the remaining class (s). Due to the complexity of soft-tissue lesions reflected in the heterogeneity of the pseudo-color images, more GMM components (n=17) were attributed to class (s), compared to the other three (c, e and h) (n=6). CONCLUSION:Combination of 2D CNN VAE and GMM achieves high classification probabilities for hard tissue-containing lesions. Automatic recognition of these classes may aid therapeutic decision-making and identifying uncrossable lesions prior to endovascular intervention.

Intraprocedural 3D real-time magnetic resonance imaging (MRI) provides a way for accurate and precise radiofrequency catheter targeting during ventricular tachycardia ablation. However, the limited data acquisition time needed to freeze cardiac motion results in highly undersampled k-space data that are challenging to reconstruct. In this work, we evaluated several deep learning (DL) based methods for real-time reconstruction of highly undersampled 3D real-time cardiac MRI. Algorithm reconstruction performance and speed were compared between classical algorithms and DL-based methods. Generative adversarial networks with attention layers in the generator were used to perform reconstructions in the image domain, which strived to balance reconstruction speed and image quality. In addition, variational networks were implemented by iterating data consistency in k-space and enforcing image smoothness via neural network-based regularization. In a preliminary study of heartbeat-resolved highly undersampled 3D cardiac MRI for 11 healthy volunteers, we observed that DL reconstruction methods provided good image quality with a significant increase in computational speed.

Cardiac parametric mapping is useful for evaluating cardiac fibrosis and edema. Parametric mapping relies on single-shot heartbeat-by-heartbeat imaging, which is susceptible to intra-shot motion during the imaging window. However, reducing the imaging window requires undersampled reconstruction techniques to preserve image fidelity and spatial resolution. The proposed approach is based on a low-rank tensor model of the multi-dimensional data, which jointly estimates spatial basis images and temporal basis time-courses from an auxiliary parallel imaging reconstruction. The tensor-estimated spatial basis is then further refined using a deep neural network, trained in a fully supervised fashion, improving the fidelity of the spatial basis using learned representations of cardiac basis functions. This two-stage spatial basis estimation will be compared against Fourier-based reconstructions and parallel imaging alone to demonstrate the sharpening and denoising properties of the deep learning-based subspace analysis.

BACKGROUND:Novel treatment strategies are needed to improve the structure and function of the myocardium post-infarction. In vitro-matured pluripotent stem cell-derived cardiomyocytes (PSC-CMs) have been shown to be a promising regenerative strategy. We hypothesized that mature PSC-CMs will have anisotropic structure and improved cell alignment when compared to immature PSC-CMs using cardiovascular magnetic resonance (CMR) in a guinea pig model of cardiac injury. METHODS:Guinea pigs (n = 16) were cryoinjured on day -10, followed by transplantation of either 108 polydimethylsiloxane (PDMS)-matured PSC-CMs (n = 6) or 108 immature tissue culture plastic (TCP)-generated PSC-CMs (n = 6) on day 0. Vehicle (sham-treated) subjects were injected with a pro-survival cocktail devoid of cells (n = 4), while healthy controls (n = 4) did not undergo cryoinjury or treatment. Animals were sacrificed on either day +14 or day +28 post-transplantation. Animals were imaged ex vivo on a 7T Bruker MRI. A 3D diffusion tensor imaging (DTI) sequence was used to quantify structure via fractional anisotropy (FA), mean diffusivity (MD), and myocyte alignment measured by the standard deviation of the transverse angle (TA). RESULTS:MD and FA of mature PDMS grafts demonstrated anisotropy was not significantly different than the healthy control hearts (MD = 1.1 ± 0.12 × 10-3 mm2/s vs 0.93 ± 0.01 × 10-3 mm2/s, p = 0.4 and FA = 0.22 ± 0.05 vs 0.26 ± 0.001, p = 0.5). Immature TCP grafts exhibited significantly higher MD than the healthy control (1.3 ± 0.08 × 10-3 mm2/s, p < 0.05) and significantly lower FA than the control (0.12 ± 0.02, p < 0.05) but were not different from mature PDMS grafts in this small cohort. TA of healthy controls showed low variability and was not significantly different than mature PDMS grafts (p = 0.4) while immature TCP grafts were significantly different (p < 0.001). DTI parameters of mature graft tissue trended toward that of the healthy myocardium, indicating the grafted cardiomyocytes may have a similar phenotype to healthy tissue. Contrast-enhanced magnetic resonance images corresponded well to histological staining, demonstrating a non-invasive method of localizing the repopulated cardiomyocytes within the scar. CONCLUSIONS:The DTI measures within graft tissue were indicative of anisotropic structure and showed greater myocyte organization compared to the scarred territory. These findings show that MRI is a valuable tool to assess the structural impacts of regenerative therapies.