Percutaneous coronary intervention (PCI) is one of the most frequently performed invasive procedures in medicine, yet it remains technically demanding, carries a long operator learning curve, exposes both patients and staff to ionising radiation, and is a recognised source of patient anxiety. Virtual reality (VR), which represent immersive, computer-generated environments experienced through a head-mounted display, has been proposed as a potential tool across the PCI pathway. This narrative review synthesises literature from the last five years on four domains in which VR has been applied to PCI: (1) operator training and skill acquisition; (2) patient-specific pre-procedural planning and rehearsal; (3) intraprocedural visualisation and guidance (where VR overlaps with augmented and mixed reality); and (4) patient-centred applications including periprocedural anxiety and pain management, education, and rehabilitation after myocardial infarction or revascularisation. The evidence is heterogeneous and dominated by small studies, feasibility reports, and a growing number of randomised trials in the patient-anxiety domain. Immersive VR reduces periprocedural anxiety and represents the most clinically mature application to date, although its analgesic effect is less consistent and it has not proven non-inferior to pharmacological sedation for pain; VR-based simulation can transfer skills to real catheterisation laboratory performance; and patient-specific VR can support planning in anatomically complex PCI. The principal barriers are the absence of robust haptic feedback, cost, limited validation, cybersickness, and a scarcity of outcome-driven randomised trials. VR is best understood not as a competitor to digital twins, physics-based simulation, or augmented/mixed reality, but as one complementary layer within a broader extended-reality ecosystem for coronary intervention. We outline priorities for future research, including standardised validation frameworks, integration with patient-specific digital twins, and adequately powered trials with clinical endpoints.

Global longitudinal strain (GLS) is a well-established prognostic marker for the early detection of cancer therapy-related cardiac dysfunction (CTRCD). We previously developed machine learning (ML) models to predict reduced GLS (Low-GLS, defined as absolute GLS < 16%) from conventional echocardiographic parameters in patients with cancer. This study aimed to externally validate the performance and generalizability of the previously developed ML models in an independent cohort. This multicenter study included patients from the Tokyo Metropolitan Tama Medical Center (TMC, n = 1484) and Shizuoka Cancer Center (SCC, n = 141) who underwent echocardiography with GLS measurements before or after anticancer chemotherapy. The exclusion criterion was left ventricular ejection fraction < 50%. Low-GLS was predicted using 25 conventional echocardiographic parameters. The previously developed ML models (Random Forest, Extra Trees, and CatBoost) were directly applied, without retraining, to an independent external validation cohort from SCC. A conventional logistic regression model was included as a reference for comparison. Model performance was assessed using the area under the receiver operating characteristic curve (AUC), and other metrics. In the internal validation cohort, the ML models demonstrated discriminative comparable performance (AUCs 0.722-0.748), whereas logistic regression achieved an AUC of 0.729. In the external validation cohort, performance was generally preserved; the Random Forest model demonstrated the highest discriminative ability (AUC 0.772), while the remaining models achieved AUCs of 0.725-0.746. Logistic regression achieved an AUC of 0.738. ML models trained on conventional echocardiographic parameters retained moderate predictive ability for reduced GLS upon external validation.

The Coronary Artery Disease Reporting and Data System (CAD-RADS) standardizes coronary CT angiography (CCTA) reporting, but not all reports contain CAD-RADS classifications. We benchmarked 54 large language model (LLM) configurations across 50 distinct models, including recent proprietary and open-weight reasoning models, for zero-shot CAD-RADS classification. We retrospectively analyzed 500 anonymized CCTA reports from four hospitals across three U.S. regions. Expert cardiovascular radiologists provided the reference standard (human inter-rater [Formula: see text]). Fifty-four model configurations (50 distinct LLMs; four configurable models tested in both thinking and non-thinking modes) spanning Llama 2 7B through recent thinking models (DeepCogito v2, Gemini 3 Pro) processed reports using identical zero-shot prompts. Performance was measured with unweighted Cohen's κ. Two LLMs met both pre-specified non-inferiority criteria ([Formula: see text] margin and entire 95% CI within the human inter-rater agreement band, [Formula: see text]-0.956): Claude 4.6 Opus ([Formula: see text], 95% CI 0.817-0.889) and the open-weight Gemma 4 31B ([Formula: see text], 95% CI 0.810-0.882), which ranked second overall. Both met the criteria at the pre-specified [Formula: see text] margin; at [Formula: see text] no model qualified. On the reports originally dictated without a CAD-RADS statement (n=343), the top models reached [Formula: see text]. Performance declined with longer thinking chains (proxy for case complexity), but thinking-mode outperformed non-thinking mode on matched difficult reports. Gemma 4 31B remained non-inferior at 3-bit quantization and fits on a 24 GB consumer GPU. Current LLMs extract the CAD-RADS stenosis severity category from unstructured CCTA reports with agreement approaching the human inter-rater band, without task-specific training. Our data shows the remarkable rise of open models. A 31B open-weight model matched top proprietary systems and runs on consumer GPUs, enabling privacy-preserving local deployment for clinical data mining.

Post-traumatic pericarditis is a subtype of post-cardiac injury syndrome (PCIS) that occurs after non-iatrogenic cardiac trauma. Post-traumatic pericarditis is rare and difficult to diagnose. We present a case of post-traumatic pericarditis progressing to transient constrictive pericarditis following a penetrating stab wound to both ventricles, illustrating the role of multimodal imaging in diagnosis and follow-up. A 22-year-old male presented with a two-day history of severe pleuritic chest pain, six weeks after surviving a penetrating stab injury to the left and right ventricles. Initial evaluation revealed no fever, a normal cardiovascular examination, and stable vital signs. Inflammatory markers were raised, and electrocardiography (ECG) showed anterolateral T-wave inversion without PR depression or ST elevation. CT pulmonary angiography demonstrated moderate pericardial effusion and mild pleural effusion. Echocardiography showed a small pericardial effusion with septal bounce, suggesting early constriction. Cardiac magnetic resonance imaging (MRI) demonstrated pericardial thickening, prominent septal bounce and flattening, and circumferential late gadolinium enhancement, in keeping with acute transient constrictive pericarditis. The patient was treated with colchicine, ibuprofen, and prednisolone. Serial follow-up MRI showed resolution of the pericardial inflammation. This case highlights post-traumatic pericarditis as a rare complication of life-threatening penetrating cardiac trauma. Serial multimodal imaging, particularly cardiac MRI, may support the diagnosis and help monitor the response to anti-inflammatory therapy.

Tetralogy of Fallot (ToF) accounts for approximately 7-10% of all congenital heart disease and is frequently complicated by right ventricular (RV) dysfunction. Three-dimensional echocardiography (3DE) offers more accurate assessment of RV volume and function, facilitating advanced evaluation of RV geometry. Building on this, a custom software tool was developed to quantify RV shape and function. This study aimed to assess the differences of 3DE derived RV shape and function parameters in ToF and healthy controls. 3DE RV datasets were acquired in 50 ToF and 50 healthy controls. RVs were segmented using commercially available software (TomTec) and further processed with custom software to quantify mean curvature (MC), area strain (AS), and longitudinal strain (LS). Measurements were performed globally and regionally across predefined RV segments. Curvature analysis identified the posterior boundary (253% [227-265] vs. 283% [260-301] vs. 294% [264-329]; for volume overload, absent overload and healthy controls, respectively; p < 0.001) and apex (464% [419-529] vs. 501% [467-541] vs. 554 [486-619]; for volume overload, absent overload and healthy controls, respectively; p < 0.001) as key regions driving RV remodeling in ToF. Septal flattening was observed in ToF with decreased concavity of the SB (p = 0.007). Global RV AS was significantly reduced in ToF (p < 0.05), with the largest impairment observed at the apex (-16% [-21;-12] vs. -14% [-17;-6] vs. -38% [-42;-28]; for volume overload, absent overload and healthy controls, respectively; p < 0.001). This in-house developed software enables comprehensive RV morphology and deformation analysis from standard 3DE imaging. Curvature analysis provides a novel and sensitive measure of RV remodeling, while AS and LS offer detailed insight into regional RV dysfunction. Improved characterization of RV remodeling may support clinical decision-making.

Accurate, early identification of transthyretin cardiac amyloidosis (ATTR-CA) is challenging yet critical for effective treatment. In this work, the modeled endpoint is scan positivity on [Formula: see text] scintigraphy, defined by the semi-quantitative Perugini visual grade (Grade 2-3 versus Grade 0-1). Two multimodal deep-learning frameworks, including Late Fusion (LF) and Cross-Modal Fusion Network (CMF-Net), are proposed to combine [Formula: see text] scintigraphy with clinical metadata for automated detection. On a curated cohort of 109 patients (62 positive, 47 negative), fusion models consistently outperformed image-only convolutional neural networks (CNNs): CMF-Net raised average accuracy by 6.9 percentage points and LF by 5.4. EfficientNet CMF-Net achieved peak accuracy 90.9% and F1-score 91.4%. Notable gains included a +30.8 percentage points sensitivity(recall) for ResNet-34 with CMF-Net and ResNet-50 LF sensitivity of 94.9%. These results show that integrating imaging and text/numeric clinical data yields superior, reproducible detection of [Formula: see text] scan positivity and may streamline scan interpretation.

We aimed to develop and assess the performance of a Machine learning (ML) model integrating common clinical features to predict arrhythmic events in patients with Hypertrophic Cardiomyopathy (HCM). Post-hoc analysis of an international multicenter registry of 531 HCM patients (49 years (IQR 35-61), 57% male) who underwent cardiac magnetic resonance (CMR). The dataset comprised clinical, echocardiographic, and CMR variables, including quantification of late gadolinium enhancement (LGE) using the + 6 SD method. The endpoint was a composite of sudden cardiac death (SCD), aborted SCD, and sustained ventricular tachycardia (VT). A total of 28 events occurred over a median follow-up of 4.1 (IQR 1.8-7.3) years. Several ML models were developed and the predictive performance of the best model was compared to the ESC HCM risk score and to the amount of LGE. The Random Forest (RF) was the most effective method showing a good performance for predicting arrhythmic events [AUC of 0.78 (95% CI: 0.76-0.82, p < 0.001)], substantially outperforming the ESC HCM risk score [AUC of 0.64 (95% CI 0.62-0.67; p < 0.001), p < 0.001 for comparison]. However, when compared to LGE alone [AUC of 0.76 (95% CI: 0.73-0.84, p < 0.001)], the RF model did not provide significant improvement in predicting the endpoint (p = 0.817 for comparison). A ML model using available clinical variables significantly outperformed the ESC HCM risk score in predicting arrhythmic events in HCM. However, its incremental value over LGE alone was weak, underscoring the strong predictive value of this imaging marker. This findings should be interpreted as exploratory and hypothesis-generating.

Left-ventricular cardiac power output (CPO), stroke-work index (LVSWI), and ventriculo-arterial coupling (VAC) capture cardiac energetics and ventricular-vascular interaction, however their bedside prognostic value in acute decompensated heart failure (ADHF) remains uncertain. To determine whether machine-learning (ML) phenotyping based on these Doppler-derived indices refines risk stratification beyond conventional echocardiography. This prospective study enrolled 500 ADHF patients with LVEF < 40%. LVSWI, CPO, and VAC were calculated from admission echocardiography. Standardized values underwent K-means clustering, and phenotype outcomes were compared using Kaplan-Meier curves and Cox regression. A traditional logistic model (EF, TAPSE, LVEDP, RVSP, RAP) was contrasted with an augmented model adding the three indices. A random forest classifier using LVSWI, CPO, and VAC predicted 6-month mortality and was internally validated. Clustering yielded two phenotypes: low-output/uncoupled (n = 262) and preserved-output/coupled (n = 238). The low-output group had higher LVEDP, RVSP, RAP, and systemic vascular resistance (all p < 0.05) despite a similar EF. Six-month mortality was 21.4% versus 11.8% (log-rank p = 0.03; HR 2.15, 95% CI 1.08-4.30). The augmented logistic model outperformed the traditional model (AUC 0.76 vs. 0.67; Brier 0.128 vs. 0.233; NRI + 11.3%; IDI + 5.6%). The random-forest model achieved AUC 0.81 (test 0.73) and ranked LVSWI as the strongest predictor. The mortality-predictive thresholds were 0.52 W for CPO and 14.4 g·min/m[2] for LVSWI. ML phenotyping based on Doppler-derived energetics identifies physiologically distinct heart failure phenotypes. Integrating CPO, LVSWI, and VAC into routine echocardiography may provide complementary physiology-based risk stratification beyond conventional echocardiographic assessment and may help guide personalized hemodynamic-targeted therapy.

Background Electrocardiographic (ECG) devices in magnetic resonance (MR) scanners have a narrow bandwidth and suffer from signal distortion, preventing measurement of QRS/ QT durations, or detection of ST/ T-wave changes during ischaemia. We integrated an external device with novel post-processing software into a 3 Tesla (T) scanner to assess feasibility and accuracy for reconstructing 12-lead ECGs prior to and during adenosine infusion. Methods Reference 12-lead ECGs were first recorded outside the MR-environment. Using three MR-safe electrode patches, in-bore ECGs were then recorded prior to and during adenosine infusion. Post-processing removed artefacts and 12-lead ECGs were reconstructed from the raw signal using a subject-specific matrix. In-bore and reference ECGs were compared using Pearson's correlation coefficient on the PQRST waveform and by manual measurements of QRS and QT durations. Adenosine-induced ischaemic changes were compared with corresponding territories on stress perfusion maps. Results In 26 participants (70.8 ± 13.0 years; 34.6% female), in-bore ECGs were recorded without additional scan time. Reconstruction took < 2 min per participant. In-bore and reference ECGs were strongly correlated (r = 0.81, interquartile range 0.72-0.88), and mean differences in QRS and QT durations were small (-12.5 ms [95% limits of agreement - 42.5, 17.5 ms], and - 5.5 ms [-54.6, 43.6 ms] respectively). Ischaemic ECG changes during adenosine predicted stress myocardial blood flow (MBF) < 1.75 ml/g/min with sensitivity 91.7% (95% confidence interval 66.7, 100.0), specificity 59.3% (39.6, 77.6), positive predictive value (PPV) 33.3% (11.4, 57.7), negative predictive value (NPV) 97.0% (90.0, 100.0) and area under curve 0.74 (0.61, 0.86). Conclusion In this proof-of-concept study, in-bore ECGs correlated closely with reference ECGs. In-bore ECGs during stress perfusion have high specificity and NPV, but low sensitivity and PPV for predicting stress MBF < 1.5 ml/g/min. This may hold value for improving R-peak detection, patient monitoring and stress perfusion map interpretation.

Left atrial (LA) strain parameters predict mortality in patients with ischaemic/non-ischaemic dilated cardiomyopathy (DCM). Left bundle branch block (LBBB) is a common conduction abnormality in those patients. To date, the impact of LBBB on LA strain parameters assessed by cardiovascular magnetic resonance imaging (cMRI) has not been studied systematically. In this retrospective single-centre study, we included 20 (10 female) controls without DCM or LBBB, 21 (11 female) patients with LBBB without DCM, 20 (7 female) patients with DCM without LBBB, and 20 (9 female) patients with DCM and LBBB. LA volumes and strain were assessed semi-automatically using the cvi42[®] and Segment[®] software. LA reservoir/conduit strain and corresponding strain rates were significantly impaired in patients with DCM (p = 0.003/0.022 respectively < 0.001/0.004) in comparison to controls. The presence of LBBB did not lead to significant changes of LA reservoir/conduit strain in controls (p = 0.919/0.391) nor in patients with DCM (p = 0.862/0.999), same was true for LA reservoir/conduit strain rate in patients with DCM (p = 0.999/0.812). LA conduit strain rate of patients with LBBB was significantly impaired in comparison to controls (p = 0.005). Left ventricular dilatation was associated with significantly impaired LA reservoir/conduit strain/strain rates. The presence of LBBB was not associated with additional impairment in LA strain/strain rate parameters in patients with ischaemic/non-ischaemic DCM, supporting the robustness of LA functional parameters in the presence of this common conduction abnormality. However, in patients with LBBB without dilatation LA conduit strain rate was significantly impaired in comparison to controls and further studies are needed to evaluate the impact of this finding.

Evidence on cardiac remodeling, left ventricular (LV) function recovery, medication use and outcomes in older ST-segment elevation myocardial infarction (STEMI) patients is limited. We examined these aspects at baseline and evolution one year post-infarct in patients aged ≥ 80 versus < 80.STEMI registry data were retrospectively analyzed. The older subgroup included patients aged ≥ 80 years at index hospitalization. Echocardiography was performed at baseline and one year after STEMI. LV remodeling was defined as > 20% relative increase in the LV end-diastolic volume. One-year cardiovascular outcomes and five-year all-cause mortality are reported.Study included 2178 patients; 132 (6.1%) were aged ≥ 80. Older patients had worse baseline LV ejection fraction (47.2 ± 9.6% versus 49.8 ± 8.2%; p < 0.001) and experienced more adverse LV remodeling (29% versus 19%; p = 0.024). Nevertheless, LV function improved similarly in both groups. Beta-blocker and renin-angiotensin-aldosterone system inhibitor use was lower (86.4% versus 94.2%; p < 0.001 and 90.2% versus 96.5%; p < 0.001, respectively) but diuretic prescription higher (18.2% versus 8.1%; p < 0.001) in older patients. One year post-infarct beta-blocker and renin-angiotensin-aldosterone system inhibitor use decreased and was similar in both groups, but older patients continued to receive more diuretics and suffered higher heart failure hospitalization rates (7.0% versus 2.6%; p = 0.006). Five-year mortality was 25.8% in older versus 4.3% in younger patients (log-rank χ2 133.2; p < 0.001).Older patients experienced more adverse LV remodeling, received less beta-blockers and renin-angiotensin-aldosterone system inhibitors but more diuretics at discharge and one year, compared to younger patients. Nevertheless, LV function recovered similarly in both groups.

To assess the value of convolutional neural network (CNN)-based denoising for the evaluation of non-calcified coronary plaques on ultrahigh-resolution (UHR) photon-counting detector (PCD) coronary CT angiography (CCTA). A dynamic phantom containing lipid-rich and fibrotic plaques with 50%-diameter stenosis (PDS) was scanned on PCD-CT under varying conditions. For in-vivo imaging, consecutive patients with non-calcified coronary plaques (NCPs) who underwent CCTA with PCD-CT were included. Image series were reconstructed using a sharp vascular kernel (Bv64) with slice thicknesses of 0.2 mm/0.4 mm, quantum iterative reconstruction (QIR) levels 3/4, and with/without CNN denoising. Phantom-based line-profile analysis yielded edge-width at half maximum (EWHM) and 10-90% rise distance as sharpness metrics. Plaque contrast-to-noise ratio (CNR) and PDS were assessed in the phantom and in patients. Two readers evaluated subjective image quality (noise, diagnostic confidence) using a four-point Likert scale. Pairwise comparisons were performed using a Bonferroni-corrected p < 0.002 for significance. Fifty-five patients (median age 74 [66.5-78] years; 19 women) with 97 NCPs were included. Phantom-based sharpness metrics remained unchanged after denoising (all pairwise p > 0.051). CNN denoising increased plaque CNR consistently in the phantom (fibrotic: 0.80-1.64 to 0.92-2.06; lipid-rich: 0.21-0.34 to 0.35-0.46) and in vivo (0.99-2.07 to 1.37-2.92, all pairwise p < 0.001). Denoising did not alter PDS values in phantom or in vivo plaques (all pairwise p > 0.006). Subjective image noise and diagnostic confidence improved across all reconstructions (all pairwise p < 0.001). CNN-based denoising of UHR PCD-CT improves image quality of non-calcified coronary plaques, while preserving sharpness and quantitative stenosis metrics.

In this exploratory diagnostic study, although our previous study demonstrated the strong prognostic value of CZT-SPECT-derived stress myocardial blood flow (sMBF) and myocardial flow reserve (MFR) in patients with ischemia and non-obstructive coronary artery disease (INOCA), their diagnostic value and optimal cut-off values for coronary microvascular dysfunction (CMD) remain unclear.

A 73-year-old asymptomatic obese woman was referred for pre-operative cardiologic evaluation before elective knee arthroplasty. Transthoracic echocardiography revealed asymmetric left ventricular hypertrophy with preserved systolic function, posterior mitral annular calcification and a large non-mobile hyperechogenic mass within the interventricular septum. Cardiac magnetic resonance confirmed asymmetric septal hypertrophy and identified septal and posterior mitral annular lesions characterized by markedly reduced native T1 and T2 values, hypoperfusion on first-pass perfusion imaging, and peripheral late gadolinium enhancement surrounding a non-enhancing core, consistent with calcification and surrounding fibrosis. Cardiac computed tomography demonstrated severe calcific degeneration of the mitral annulus extending into the interventricular septum and adjacent myocardium. Secondary metabolic, inflammatory, ischemic and infectious causes of myocardial calcification were excluded. This case highlights an unusual presentation of caseous mitral annular degeneration with intramyocardial extension mimicking a cardiac mass.