Suchá, Dominika

Introduction: This study aimed to develop and evaluate a deep learning (DL) deblurring algorithm for computed tomography (CT). Phantom testing assessed whether DL deblurring could improve CT image quality and coronary artery calcium (CAC) detectability. Materials and Methods: The algorithm consisted of 2 sequential convolutional neural networks (CNNs): one for denoising and one for deblurring. After deblurring, the estimated noise from the first model was reintroduced to preserve image texture. The CNN denoising model was trained to map high-noise images to routine low-noise ones. The CNN deblurring model was trained using natural images—blurred inputs paired with sharp targets. To assess performance, an anthropomorphic phantom with 100 small calcifications was scanned on a CT system using a CAC scoring protocol. Data were acquired at clinical and high doses, then reconstructed with and without DL deblurring, using 25% of the maximum denoising strength. Detectability was defined as the ability to calculate an Agatston score (≥3 adjacent voxels >130 HU). Results: At high dose, CAC detectability increased from 39% (standard reconstruction) to 48% with deblurring. Similar detectability (39%) was achieved at the routine dose with deblurring, representing a 6% improvement compared with the original routine dose data. Discussion: Quantitative results from phantom testing demonstrate that DL deblurring can improve image quality, based on a clinically relevant, task-based CAC detection metric at clinical dose levels.

Articles published in Radiology: Cardiothoracic Imaging in 2025 highlight advances in cardiothoracic imaging through technical innovation, quantitative biomarkers, and artificial intelligence applications that support improved diagnosis, risk stratification, and patient-centered management across cardiac, vascular, and thoracic disease.

Abstract Introduction Cardiovascular magnetic resonance (CMR) imaging enables precise, non-invasive assessment of cardiac structure and function, yet the clinical utility of CMR metrics is limited by the lack of comprehensive, population-specific reference values. By directly reanalyzing data from a large, multi-ethnic cohort rather than relying solely on pooled literature, could enable the establishment of standardized reference ranges and severity grading thresholds. Aim To provide comprehensive, age-, sex-, and ethnicity-stratified CMR reference ranges and severity grading thresholds for key ventricular and atrial parameters. Methods We analyzed over 9,000 CMR scans from the Healthy Hearts Consortium (HHC), comprising six international cohorts. All individuals were free from overt cardiovascular disease and major comorbidities, ensuring a verified healthy sample. Two automated segmentation software solutions—cvi42 (Circle Cardiovascular Imaging) and suiteHEART (Neosoft)—were used to quantify left and right ventricular volumes, myocardial mass, and atrial volumes in standard cine stacks. All endorsed segmentation modes were applied. Quality control protocol combined expert visual review and statistical outlier analysis. We derived 95% prediction intervals (PI) to define the normal range, with additional severity thresholds for mild (95–99.73% PI), moderate (99.73–99.99% PI), and severe (>99.99% PI) deviation from the reference range. Results The final dataset comprised of 9,059 verified healthy adults aged 18–83 years (mean 61 ± 13 years), with a near-equal sex distribution (51% women). Demographics included White (81.6%), South Asian (5.6%), Mixed/Other (5.3%), Black (3.8%), and Chinese (3.7%) participants. Results demonstrated that sex, age, and ethnicity strongly influenced cardiac indices. Men had higher indexed LV and RV volumes and greater myocardial mass than women. Black participants presented the highest indexed LV myocardial mass across age groups, whereas Chinese participants had comparatively lower myocardial mass values (Figure 1). In contrast, atrial volumes remained relatively stable across age categories. Comparative analysis showed minor differences between the two software tools, particularly for atrial parameters; however, derived ejection fractions remained consistent. We established age-, sex-, and ethnicity-stratified reference ranges for key CMR parameters. To facilitate clinical interpretation, we defined severity thresholds corresponding to mild, moderate, and severe abnormalities. Conclusion This large-scale, international effort addresses prior limitations in CMR interpretation by offering robust, contemporary reference ranges and severity grading applicable to diverse populations and imaging protocols. Our severity grading thresholds support consistency in clinical reporting and decision-making, enabling clearer identification of early subclinical changes and more accurate monitoring of disease progression.Subset of CMR value ranges in females

The field of cardiothoracic imaging is expanding rapidly with new techniques that enable greater diagnostic accuracy and more personalized patient management, ultimately improving outcomes and enhancing clinical decision-making.

Fetal cardiovascular magnetic resonance imaging (MRI) has emerged as a complementary modality for prenatal imaging in suspected congenital heart disease. Ongoing technical improvements extend the potential clinical value of fetal cardiovascular MRI. Ascertaining equivocal prenatal diagnostics obtained with ultrasonography allows for appropriate parental counseling and planning of postnatal surgery. This work summarizes current acquisition techniques and clinical applications of fetal cardiovascular MRI in the prenatal diagnosis and follow‐up of fetuses with congenital heart disease.Level of Evidence3Technical EfficacyStage 3