Radiology: Cardiothoracic Imaging Highlights 2024

Abstract

Radiology: Cardiothoracic Imaging publishes research, technical developments, and reviews related to cardiac, vascular, and thoracic imaging. The current review article, led by the Radiology: Cardiothoracic Imaging trainee editorial board, highlights the most impactful articles published in the journal between November 2023 and October 2024. The review encompasses various aspects of cardiac, vascular, and thoracic imaging related to coronary artery disease, cardiac MRI, valvular imaging, congenital and inherited heart diseases, thoracic imaging, lung cancer, artificial intelligence, and health services research. Key highlights include the role of CT fractional flow reserve analysis to guide patient management, the role of MRI elastography in identifying age-related myocardial stiffness associated with increased risk of heart failure, review of MRI in patients with cardiovascular implantable electronic devices and fractured or abandoned leads, imaging of mitral annular disjunction, specificity of the Lung Imaging Reporting and Data System version 2022 for detecting malignant airway nodules, and a radiomics-based reinforcement learning model to analyze serial low-dose CT scans in lung cancer screening. Ongoing research and future directions include artificial intelligence tools for applications such as plaque quantification using coronary CT angiography and growing understanding of the interconnectedness of environmental sustainability and cardiovascular imaging.

Keywords: CT, MRI, CT-Coronary Angiography, Cardiac, Pulmonary, Coronary Arteries, Heart, Lung, Mediastinum, Mitral Valve, Aortic Valve, Artificial Intelligence

© RSNA, 2025

Summary

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.

Essentials

• ■ CT fractional flow reserve analysis has decreased rates of invasive intervention and was feasible even without heart rate control.

• ■ Left ventricular global longitudinal peak strain and extracellular volume provide incremental value for predicting sustained ventricular arrhythmias in patients with arrhythmogenic right ventricular cardiomyopathy.

• ■ Classification of airways nodules based on the Lung Imaging and Reporting Data System (Lung-RADS) version 2022 showed higher specificity compared with Lung-RADS version 1.1, decreasing the number of false-positive screening CT examinations while still identifying malignant airway nodules.

• ■ Artificial intelligence–based coronary stenosis quantification allows accurate identification of moderate and severe coronary artery stenosis.

• ■ Many strategies available to improve environmental sustainability in cardiovascular imaging, such as turning imaging equipment off when not in use or abbreviating protocols, should be integrated in daily practice given the high impact of climate change on cardiovascular health.

Introduction

Cardiothoracic imaging is a field that is constantly evolving, with emerging research advancing our understanding of cardiovascular and respiratory diseases. Building on prior review articles highlighting key publications in 2022 and 2023 (1,2), the current review focuses on most recent advancements in cardiovascular and thoracic imaging published in Radiology: Cardiothoracic Imaging. This manuscript was led by the Radiology: Cardiothoracic Imaging trainee editorial board.

Article selection for inclusion followed a systematic approach. All manuscripts published between November 2022 and October 2023 were considered, with the exception of those categorized as Images in Cardiothoracic Imaging, which are evaluated in a separate trainee editorial board–led publication (3,4). All eligible articles were initially ranked within four separate categories based on data collected in December 2024: total number of downloads, number of downloads per month since online publication date, the Altmetric score, and total number of citations. The top 10% of articles ranked in each of the four categories were included in the review. These metrics tend to favor articles published earlier in the time period. Therefore, trainee editorial board members also included additional articles based on timeliness, novelty, and reader interest. This approach aimed for a balance between quantitative ranking and qualitative assessment. Articles were initially reviewed by trainee editorial board members (R.C. and A.M.), with additional editorial input from Radiology: Cardiothoracic Imaging editors.

Of the 67 total articles published between November 2023 and October 2024, 27 were selected for inclusion: 22 articles based on the top 10% of each of the four initial rank lists after removing duplicates ranked within the top 10% in multiple categories and five additional articles based on timeliness, novelty, and reader interest. Included articles were grouped into one of nine categories: coronary artery disease (CAD), cardiac MRI, congenital and inherited heart diseases, valvular diseases, thoracic imaging, lung cancer imaging, artificial intelligence (AI), health services and sustainability, and future perspectives. All authors of articles published in Radiology: Cardiothoracic Imaging are commended for making a substantial contribution to the journal, its readers, and the field of cardiothoracic imaging.

Coronary Artery Disease

Coronary CT angiography (CCTA) is a first-line imaging modality to evaluate CAD in patients with stable chest pain; however, specificity for obstructive disease is limited. CT fractional flow reserve (CT-FFR) is a physiologic simulation technique that uses routine CCTA data to mathematically model coronary flow, pressure, and resistance (5). Randhawa et al (6) evaluated CT-FFR in a large single-center retrospective cohort of patients undergoing dual-source CCTA without heart rate control. Overall, 292 of 2985 patients (9.7%) were referred for CT-FFR analysis, mainly for Coronary Artery Disease Reporting and Data System 3 lesions (74%). Only two studies (0.7%) were rejected from CT-FFR analysis. The observational study design and selection bias limit generalizability, but CT-FFR showed potential to reduce unnecessary invasive coronary angiography. Among patients with significant stenosis at CCTA, patients who underwent CT-FFR analysis had lower rates of invasive coronary angiography (25.5% vs 74.5%; P = .04) and percutaneous coronary intervention (21.1% vs 78.9%, P = .05) compared with those without CT-FFR analysis.

CCTA with CT-FFR provides anatomic and hemodynamic information, and coronary volume–to–myocardial mass ratio may be used to indicate coronary supply–myocardial demand mismatch. In a subset of patients from the ADVANCE registry (ClinicalTrials.gov: NCT02499679), Holmes et al (7) investigated the relationship between smoking status and coronary volume–to–myocardial mass ratio in 2874 patients with CAD with coronary stenosis of 30% or greater at CCTA. Both current and former smokers had lower coronary volume–to–myocardial mass ratio ± SD than never-smokers (current smokers, 24.1 mm³/g ± 7.9; former smokers, 24.9 mm³/g ± 7.1; never-smokers, 25.8 mm³/g ± 7.4; P < .001 [unadjusted] and P = .002 [unadjusted], respectively), mostly driven by higher myocardial mass (Fig 1). Stenosis of 50% or greater and diabetes were also associated with low coronary volume–to–myocardial mass ratio. Further research is needed to determine the role of coronary volume–to–myocardial mass ratio in diagnosis of microvascular dysfunction.

Figure 1:

Sample CT fractional flow reserve models and calculated coronary volume–to–myocardial mass ratio (V/M, in cubic millimeters per gram) in current smokers and never-smokers demonstrate low and high V/M, respectively. (Reprinted, with permission, from reference 7.)

People living with HIV are at increased risk of cardiovascular disease and adverse events, even without classic risk factors. Abd-Elmoniem et al (8) prospectively evaluated 74 adults with HIV without cardiovascular disease and 25 matched controls using 3.0-T MRI to measure proximal right coronary artery vessel wall thickness. Vessel wall thickness ± SD was significantly higher in people living with HIV compared with controls (1.47 mm ± 0.22 vs 1.34 mm ± 0.18; P = .006). Among people living with HIV, higher coronary artery vessel wall thickness was associated with a lower ratio between the left ventricular (LV) filling peak blood flow velocity in early diastole to that in late diastole (P < .001) and higher LV mass index (P = .03), indicating restricted diastolic function. These results suggest that MRI-derived vessel wall thickness may be useful for early recognition of coronary artery pathology in people living with HIV.

Cardiac MRI

There are limited data on the safety of MRI in patients with cardiovascular implantable electronic devices with fractured or abandoned leads (9). Greenhill et al (10) evaluated the effect of lead length and lead orientation in patients with cardiovascular implantable electronic devices and lead fragments or abandoned leads. They evaluated 80 patients who underwent a total of 107 1.5-T MRI examinations of various body regions. There were no clinically significant adverse outcomes within 30 days of imaging. There were no reported deaths, clinically significant arrhythmias, or adverse clinical events within 30 days of MRI examination. However, three of 67 patients with abandoned leads had a significant change in the composite of capture threshold, sensing, or lead impedance between before MRI and after MRI examination.

Arani et al (11) evaluated the influence of age and sex on LV shear stiffness using MR elastography. MR elastography uses extrinsic mechanical vibrations on the chest wall to generate shear waves in the myocardium allowing quantification of myocardial shear stiffness (12). This prospective study included 109 healthy volunteers and found that myocardial shear stiffness increased with age in women (age slope = 0.03 kPa/year ± 0.01; P = .009) but not men (age slope = 0.008 kPa/year ± 0.009, P = .38) (Fig 2).

Figure 2:

The range of short-axis left ventricular stiffness maps observed with MRE in younger and older healthy volunteers. The regions of interest used for the cardiac MRE measurements are outlined in white in the short-axis MRI magnitude images and red in each corresponding elastogram. The ejection fraction (EF) was not significantly correlated with myocardial stiffness. There was no use of contrast media for any images. MRE = MR elastography. (Reprinted, with permission, from reference 12.)

Hopman et al (13) compared pulse sequence image-navigated three-dimensional late gadolinium enhancement (LGE) cardiac MRI to traditional diaphragm-navigated three-dimensional LGE cardiac MRI for evaluating left atrial LGE prospectively in 26 participants with atrial fibrillation (Fig 3). Acquisition time and LGE quantification were significantly lower for the image-navigated three-dimensional LGE sequence compared with diaphragm-navigated three-dimensional LGE (4.9 minutes ± 1.1 vs 12 minutes ± 4; P < .001 and 12% ± 8 vs 20% ± 12; P < .001). Reviewers selected diaphragm-navigated three-dimensional LGE as the preferred strategy in 65% of cases, suggesting that further study and refinement are needed before clinical integration.

Figure 3:

Case example of fibrosis overlap at image intensity ratio (IIR) threshold of 1.2. Graphic shows a visual representation of spatial correspondence between detected left atrial (LA) fibrosis in the diaphragm-navigated (dNAV) scan, compared with the detected LA fibrosis in the image-navigated (iNAV) scan. Top row shows the iNAV and dNAV healthy myocardium (blue) and fibrosis (red) using an IIR threshold of 1.2. The detected fibrosis percentage is annotated in both cases. Bottom row shows the spatial correspondence of fibrosis detection, showing true negatives (blue), false negatives (red), false positives (green), and true positives (yellow). The Dice similarity score recorded for this case was 0.42. In all images, white corresponds to the pulmonary vein cutoffs or LA appendage cutoff. 3D = three-dimensional. (Reprinted, with permission, from reference 13.)

Valvular Disease

In a prospective study, Lee et al (14) identified hypoattenuating intra-annular material of the bioprosthetic aortic valve at CT in 18 of 23 (78.3%) participants with failed surgical bioprosthetic aortic valves undergoing valve-in-valve transcatheter aortic valve replacement. Participants with hypoattenuating intra-annular material had a larger difference between the true valve internal diameter and the internal diameter at CT compared with those without (3.70 mm ± 1.09 vs 1.40 mm ± 0.73; P < .001) because the internal diameter at CT accounts for the presence of hypoattenuating intra-annular material. CT-derived internal diameter correlated with postprocedural indexed effective orifice area in participants with intra-annular positioned valves (ρ = 0.69, P = .004), suggesting that supra-annular positioning may be a better option in this population.

Mitral annular disjunction and mitral valve prolapse are associated and are increasingly recognized at imaging, with debate regarding clinical significance. Gulati et al (15) provide a comprehensive review of mitral annular disjunction, including association with risk of ventricular arrhythmias and sudden cardiac death (16). They identified high-risk features including mitral annular disjunction greater than 5 mm, paradoxical annular curling, and LGE of the peri-annular basal LV or papillary muscles (Fig 4, Table 1). Akyea et al (17) used unsupervised machine learning in 474 patients with mitral valve prolapse without hemodynamically significant mitral regurgitation or LV dysfunction. Two phenotypic clusters were identified based primarily on cardiac MRI features. Patients in the cluster with more severe mitral valve degeneration, cardiac chamber remodeling, and LGE at MRI were at higher risk of the composite end point of sustained ventricular tachycardia, aborted sudden cardiac death, or unexplained syncope (hazard ratio: 3.79; 95% CI: 1.19, 12.12; P = .02) after adjustment for LGE extent.

Figure 4:

Images in a 24-year-old asymptomatic female patient with mitral valve prolapse diagnosed at 3 years of age. (A) Her electrocardiogram showed normal sinus rhythm. A three-chamber view at echocardiography showed a 10-mm gap between the mitral valve hinge point and left ventricular myocardium (black arrow). On the (B) two-chamber and (C) three-chamber views of cine steady-state free precession cardiac MR images, this gap was confirmed in the end-systolic phase and measured 6 mm (black arrow). (Reprinted, with permission, from reference 15.)

Table 1:

Studies with More than 50 Participants that Linked Cardiac Arrhythmia, MAD, and Cardiac MRI or Echocardiography

Congenital and Inherited Heart Diseases

Arrhythmogenic cardiomyopathy is an inherited cardiomyopathy associated with risk of arrhythmia and sudden cardiac death, initially thought to primarily affect the right ventricle. However, there is increasing recognition of LV-dominant and biventricular phenotypes. A retrospective study by Lu and Cao et al (18) explored the role and incremental value of cardiac MRI feature tracking and T1 mapping in predicting sustained ventricular arrhythmias beyond traditional risk scores in 91 patients with arrhythmogenic cardiomyopathy (19). Worsening LV and right ventricular global longitudinal peak strain (1% change; hazard ratio for LV: 1.14; 95% CI: 1.06, 1.223; P = .001; hazard ratio for right ventricular: 1.09; 95% CI: 1.02, 1.16; P = .02) and increased extracellular volume fraction (1% change; hazard ratio: 1.13, 95% CI: 1.08, 1.18; P < .001) were associated with increased risk of sustained ventricular arrhythmias, respectively, after adjustment for the traditional arrhythmogenic cardiomyopathy risk score. Adding both biventricular global longitudinal peak strain and extracellular volume fraction to the risk score had incremental value for predicting sustained ventricular arrhythmia (area under the receiver operating characteristic curve: 0.73 vs 0.65; P < .001) (Fig 5).

Figure 5:

Images of cardiac MRI biventricular global longitudinal peak strain (GLS) analysis and extracellular volume fraction (ECV) map. A 42-year-old female patient diagnosed with arrhythmogenic right ventricular cardiomyopathy based on the revised 2010 Task Force criteria who experienced sustained ventricular arrhythmias (VA) as a clinical end point during follow-up. (A) Left ventricular (LV) GLS (−12.97%) in long-axis planes (four-, three-, two-chamber view) and right ventricular (RV) GLS (−11.16%) in long-axis plane (four-chamber view). (B) Native T1, post-T1, and ECV (31.2%) map in the midventricular section of short-axis plane. (Reprinted, with permission, from reference 18.)

Cardiac involvement is the leading cause of death in Fabry disease, characterized by LV hypertrophy and myocardial fibrosis (20,21). A systematic review and meta-analysis by Figliozzi and Kollia et al (22) evaluated the effect of enzyme replacement therapy on cardiac MRI parameters in patients with Fabry disease and included 11 studies for a total of 445 patients. Between baseline and follow-up cardiac MRI, enzyme replacement therapy was associated with reduced LV maximum wall thickness (mean difference, −1 mm; 95% CI: −2, −0.02; six studies, 151 patients, I² = 90%) and increased LGE extent (mean difference, 1%; 95% CI: 1, 1; three studies, 114 patients, I² = 85%) with substantial heterogeneity.

Patients with congenital heart disease and single-ventricle physiology are at risk for heart failure (23,24). In a retrospective review of 43 patients with single-ventricle physiology who underwent noncontrast T2-weighted lymphatic MRI, Kelly et al (25) found that lymphatic abnormalities progressed in 19 individuals after Fontan completion. Patients with progressive lymphatic abnormalities had longer hospitalization after Fontan completion (median time, 13 days; IQR, 9–25 vs 26 days; IQR, 18–30; P = .03) and were more likely to develop chylothorax (12% [three of 24] vs 75% [six of eight]; P < .01) and/or protein-losing enteropathy (0% [0 of 24] vs 38% [three of eight]; P < .01) during the median 8-years follow-up period (Fig 6).

Figure 6:

Maximal intensity projections of noncontrast lymphatic imaging in two individuals at pre-Fontan and Fontan stages. (A) Image in a 2-year-old girl at pre-Fontan stage and (B) image 2 years later at Fontan stage displaying no notable changes. The thoracic duct is visible and displaying a similar morphology on both images (arrows). The patient did not develop complications in the postoperative period. (C) Image at pre-Fontan staging in a 2-year-old girl displaying type 3 lymphatic classification (more mediastinal abnormal signal intensity visible on multisection image). (D) Image 2 months after Fontan completion in the same patient. The terminal part of the thoracic duct can be visualized on both images (arrowheads). A progression in abnormal signal intensity is especially apparent in the supraclavicular and mediastinal regions (arrows), imaging now categorized as type 4. The postoperative period was characterized by readmission for pleural effusion, chylothorax, and lymphatic intervention. (Reprinted, with permission, from reference 25.)

Thoracic Imaging

Chest CT has a central role in the diagnosis, treatment planning, and follow-up of interstitial lung diseases. Chelala et al (26) evaluated the diagnostic performance of the American Thoracic Society, Japanese Respiratory Society, and Asociacion Latinoamericana del Torax versus American College of Chest Physicians imaging classifications for hypersensitivity pneumonitis in a retrospective study of 297 patients (67% with hypersensitivity pneumonitis, 16% with connective tissue associated interstitial lung disease, and 16% with idiopathic pulmonary fibrosis). Diagnostic performance varied based on disease prevalence, with a 21% discordance rate between the two systems. The American Thoracic Society, Japanese Respiratory Society, and Asociacion Latinoamericana del Torax classification had a higher accuracy assuming low (10%) disease prevalence (92.3% vs 87.6%), whereas the American College of Chest Physicians classification had higher accuracy assuming a high (50%) prevalence (81.7% vs 79.7%).

Shah et al (27) reviewed the imaging features of non–idiopathic pulmonary fibrosis–pattern fibrosis and implications for therapy (Table 2). The term progressive pulmonary fibrosis is used to describe non–idiopathic pulmonary fibrosis lung disease that demonstrates clinical, physiologic, and/or radiologic progression over the course of a year and may be treated with antifibrotic therapy, thus impacting management and potentially survival (Fig 7).

Table 2:

Findings in Fibrotic Lung Disease Suggesting an Alternative Diagnosis to a UIP Pattern and Their Associated Differential Diagnosis

Figure 7:

Texture-based CT quantification demonstrates progressive disease in a 68-year-old male patient with hypersensitivity pneumonitis. (A) Coronal CT image with texture analysis overlay reveals a data-driven texture analysis score of 26%. (B) Coronal CT image with texture analysis overlay at 3-year follow-up demonstrates progression, with a data-driven texture analysis score of 39%. (Image courtesy of David Lynch, MD, and Stephen M. Humphries, PhD, Quantitative Imaging Laboratory, National Jewish Health.) (Reprinted, with permission, from reference 27.)

Congenital or acquired dynamic instability of the central airways is often underestimated and underdiagnosed (28,29). Bischoff et al (30) explored the role of low-dose four-dimensional CT for tracheal collapsibility quantification and compared visual and quantitative four-dimensional CT-based assessment with paired inspiratory-expiratory CT, bronchoscopy, and spirometry in a retrospective study of 52 patients. Visual assessment of four-dimensional CT detected more patients with tracheal collapsibility of 50% or greater compared with both paired CT (concordance 41%, P < .001) and bronchoscopy (concordance 74%, P = .39), highlighting the feasibility and sensitivity of this technique.

Lung Cancer Imaging

Liu et al (31) evaluated the clinicopathologic characteristics and prognosis of patients with clinical stage IA lung adenocarcinoma with atypical solid nodules at chest CT who underwent resection. Overall, 49 of 254 patients had atypical solid nodules at CT, defined as a solid nodule with multiple pseudocavities (≥3) with a sievelike appearance in both lung and mediastinal windows compared with typical solid nodules. Patients with atypical solid nodules had lower incidence of lymph node metastasis (6.4% vs 24.7%, P = .009) and had longer average disease-free survival compared with patients with typical solid nodules (P < .001) (Fig 8).

Figure 8:

CT and pathologic findings of atypical solid nodules (ASNs). (A) Images in a 63-year-old man with acinar predominant invasive adenocarcinoma (IAC) presenting as ASN. The first two CT images depict the ASN, captured in the lung and mediastinal windows, respectively, using nonenhanced CT scans. The third image shows a pathologic specimen of ASN (hematoxylin-eosin stain; magnification, ×40). (B) Images in a 58-year-old man with papillary predominant IAC presenting as ASN. The first two CT images depict the ASN, captured in the lung and mediastinal windows, respectively, using nonenhanced CT scans. The third image shows a pathologic specimen of ASN (hematoxylin-eosin stain; magnification, ×40). (C) Images in a 58-year-old man with micropapillary predominant IAC presenting as ASN. The first two CT images depict the ASN, captured in the lung and mediastinal windows, respectively, using contrast-enhanced CT scans. The third image shows a pathologic specimen of ASN (hematoxylin-eosin stain; magnification, ×40). (D) Images in a 72-year-old man with invasive mucinous adenocarcinoma presenting as ASN. The first two CT images depict the ASN, captured in the lung and mediastinal windows, respectively, using nonenhanced CT scans. The third image shows a pathologic specimen of ASN (hematoxylin-eosin stain; magnification, ×40). (Reprinted, with permission, from reference 31.)

DeSimone et al (32) compared Lung Imaging Reporting and Data System (Lung-RADS) version 1.1 with Lung-RADS version 2022 for airway nodules detected at lung cancer screening CT examinations. In a retrospective analysis of 174 patients with a reported airway or endobronchial nodule at screening CT examination, 163 (94%) were benign, and 11 (6%) were malignant. Version 2022 had higher specificity compared with version 1.1 (82% vs 50%), with similar sensitivity (91% for both). Imaging features associated with benign nodules included location in the trachea or mainstem bronchi, multiplicity, nonobstructive morphologies, dependent portions of airway, internal air, and fluid attenuation.

Oh et al (33) evaluated preoperative risk factors in a retrospective analysis of 366 patients with pathologic stage IIIA N2 non–small cell lung cancer who underwent upfront surgery. Clinical nodal categories (hazard ratio for cN1 vs cN0: 1.91; 95% CI: 1.11, 3.27; hazard ratio for CN2a2 vs cN0: 1.89; 95% CI: 1.13, 2.18; and hazard ratio for cN2b vs cN0: 2.02; 95% CI: 1.07, 3.80) and lymph node size at N1 station (hazard ratio: 1.75; 95% CI: 1.12, 2.71) were independent prognostic factors for overall survival. This suggests that reporting single- versus multistation involvement of N2 disease and maximum size of metastatic lymph nodes at preoperative chest CT provides useful prognostic information (34,35).

Artificial Intelligence

AI offers promising tools for many steps of the cardiovascular and thoracic imaging workflow, from patient and test selection to image acquisition, reconstruction, interpretation, prognostication, and reporting (36). Wang et al (37) evaluated the feasibility of leveraging serial low-dose chest CT scans to develop a radiomics-based reinforcement learning model for improving early diagnosis of lung cancer at baseline screening using data from participants randomly selected from the National Lung Screening Trial. The radiomics-based reinforcement learning model uses advanced imaging features and sequential decision-making algorithms to analyze serial low-dose CT scans and had a higher area under the receiver operating characteristic curve (0.88; 95% CI: 0.85, 0.91) than the Brock model (0.84; 95% CI: 0.81, 0.88; P = .02), a widely used clinical tool for lung cancer risk assessment. This highlights the potential for AI-based tools to improve early diagnosis and risk stratification for lung cancer at baseline CT screening.

Dundas et al (38) assessed the performance of a commercially available AI-based tool by comparing the quantified stenosis severity at coronary CCTA with invasive coronary angiography as the reference standard in a secondary, post hoc analysis that included 120 participants from three large trials. Quantification using the AI-based tool had areas under the receiver operating characteristic curve to predict degree of stenosis of 70% or greater on a per-vessel basis of 0.93 (95% CI: 0.89, 0.97; P < .001) and a per-patient basis of 0.88 (95% CI: 0.81, 0.94; P < .001) (Fig 9).

Figure 9:

Examples of discordant cases between quantitative coronary angiography (QCA) and artificial intelligence–based coronary artery stenosis quantification (AI-CSQ). (A–D) Images demonstrate a vessel with a severe stenosis. (A) Image from invasive QCA and (B) CT image of the vessel, with luminal cross-section inset, indicate the location of the large side branch adjacent to the stenosis (red arrowhead). (C) The straightened vessel multiplanar reformat indicates the AI-CSQ determination of the lumen (blue lines), with the arrowhead demonstrating its erroneous inclusion of some of the same side branch in the main vessel lumen. (D) The resulting AI-CSQ three-dimensional anatomic model underestimates the severe stenosis. (E–H) Images demonstrate a different vessel with a severe stenosis (red arrowheads) by using (E) QCA, with poor image quality at (F) CT because of motion, leading to AI-CSQ modeling the stenosis (G, H) but underestimating its severity. (Reprinted, with permission, from reference 38.)

Zhang et al (39) integrated radiomics features from cardiac MR cine images obtained on multivendor 3.0-T scanners with clinical and standard cardiac MRI predictors to identify patients with hypertrophic cardiomyopathy at high risk for heart failure. The combined model with radiomics features and clinical and standard cardiac MRI parameters had the highest 1- and 3-year areas under the receiver operating characteristic curve of 0.82 and 0.77 in the validation set. Further studies are needed to evaluate diagnostic performance in larger cohorts and better understand how AI tools can be integrated into clinical practice to facilitate faster, more reproducible, and more accurate reporting.

Health Services and Sustainability

Coughlan et al (40) summarized the most relevant cardiovascular imaging articles published in 2023, focusing on CT and MRI. Key trends include advances in CT-FFR analysis in patients with CAD, utility of photon-counting detector CT in coronary stenosis quantification, and the evolving role of inflammatory pericoronary fat evaluation as an additional prognostic imaging biomarker of high-risk coronary plaque in predicting subsequent events (Fig 10).

Figure 10:

Kuneman et al evaluated the association between pericoronary adipose tissue (PCAT) and culprit plaques by using CT. Patients with acute coronary syndrome (ACS) within 2 years after CT were identified and compared with controls. The PCAT at CT was defined as tissue with attenuation between −190 and −30 HU and within a radial distance from the vessel wall equal to the vessel diameter. (A) Multiplanar reconstructions of CT scans show mixed plaques in the proximal (right) and middle (left) left anterior descending coronary artery, with surrounding PCAT (orange-yellow colored areas) across a precursor of a culprit lesion in a patient who developed an ACS (left) and across a lesion in a patient with stable coronary artery disease (CAD) (right). (B) Mean PCAT attenuation across precursors of culprit lesions versus nonculprit lesions in patients who developed ACS versus lesions in patients with stable CAD. The study highlights the evolving role of inflammatory pericoronary fat evaluation as an additional prognostic imaging biomarker of high-risk coronary plaque in predicting subsequent events. (Adapted, with permission under a CC BY-NC-ND 4.0 license, from reference 48.)

Gunasekaran et al (41) highlighted the impact of climate change on cardiovascular health, discussed the environmental impact of cardiovascular imaging, and described opportunities to improve environmental sustainability of cardiovascular imaging (Fig 11). Strategies to improve environmental sustainability include prioritizing imaging tests with lower emissions when more than one test is appropriate, turning equipment off when not in use, recycling and reusing to reduce waste, and switching from single-use to reusable supplies (Table 3) (42). The health impacts of environmental exposures will be particularly relevant to cardiothoracic imagers given the exacerbation of cardiovascular and respiratory disease and increased medical imaging utilization related to extreme heat and poor air quality (43).

Figure 11:

Summary of opportunities and actions to improve environmental sustainability in cardiovascular imaging. (Reprinted, with permission, from reference 41.)

Table 3:

Opportunities to Address Climate Change in Cardiovascular Imaging with Rationale and Impact

In a retrospective analysis of data from the European Society of Cardiovascular Radiology MR/CT Registry, Moser et al (44) evaluated the prevalence of clinically relevant extracardiac findings at cardiac CT and MRI and the relationship with examination indications and patient characteristics. The study included data from 208 506 cardiac CT examinations and 228 462 cardiac MRI examinations. The authors found that the prevalence of clinically relevant extracardiac findings was 3.28% at cardiac CT and 1.50% at cardiac MRI. Older patient age and examination indication were associated with higher prevalence of extracardiac findings.

Future Perspectives and Novel Imaging Developments

Capaldi et al (45) prospectively studied the responsiveness of proton (¹H) MRI-derived specific ventilation to bronchodilator therapy in 27 participants with severe asthma. ¹H MRI-derived specific ventilation improved after bronchodilator administration (from 0.07 ± 0.04 to 0.11 ± 0.04; P < .001), achieving levels similar to those of healthy patients (Fig 12).

Figure 12:

Tidal breathing proton (¹H) MRI-derived specific ventilation (SV) images (heatmap) coregistered to anatomic ¹H MRI scans (grayscale) in a healthy control and two representative participants with severe asthma. (Top) Images in a 46-year-old female healthy control participant with MRI SV of 0.13. (Middle) Images in a 45-year-old female participant with asthma and type 2 low inflammatory biomarkers (T2-low) (fraction of exhaled nitric oxide [FeNO], 52 ppb; no sputum eosinophil-free granules [EFGs]; 0% sputum eosinophil percentage; 0 cells/μL [0 cells × 10⁹/L] blood eosinophil count) with prebronchodilator (pre-BD) MRI SV, postbronchodilator (post-BD) MRI SV, and ΔMRI SV of 0.03, 0.17, and 0.14, respectively. (Bottom) Images in a 63-year-old female participant with asthma and type 2 high inflammatory biomarkers (T2-high) (FeNO, 34 ppb; many sputa EFGs; 11.5% sputum eosinophil percentage; 900 cells/μL [0.9 cells × 10⁹/L] blood eosinophil count) with pre-BD MRI SV, post-BD MRI SV, and ΔMRI SV of 0.078, 0.073, and −0.005, respectively. (Reprinted, with permission, from reference 45.)

Dohna et al (46) evaluated phase-resolved functional lung MRI, a noncontrast, free-breathing technique to quantify regional ventilation and perfusion, in 23 participants with cystic fibrosis before and 8–16 weeks after elexacaftor-tezacaftor-ivacaftor therapy. Ventilation defect percentage of regional ventilation decreased from 18% to 9% (P = .003) and perfusion defect percentage from 26% to 19% (P = .002), highlighting the feasibility of this MRI technique for quantitative assessment of perfusion and ventilation changes in response to therapy (Fig 13).

Figure 13:

Phase-resolved functional lung (PREFUL) MRI ventilation and perfusion maps in an 18-year-old male participant with cystic fibrosis at baseline (quantified perfusion [QQuant]: 47 mL/min/100 mL, flow-volume loop correlation metrics [FVL_CM]: 92%, ventilation defect percentage of FVL correlation metrics [VDP_FVL-CM]: 14%, ventilation perfusion match [VQM] healthy: 70%, VQM defect: 9%) and 13 weeks after initiation of treatment with elexacaftor-tezacaftor-ivacaftor (ETI) (QQuant: 57 mL/min/100 mL, FVL_CM: 98%, VDP_FVL-CM: 4%, VQM healthy: 79%, VQM defect: 1%). Contrast-free PREFUL maps allow for visual detection of ventilation and perfusion defects and their resolution under treatment with ETI. Note improved perfusion (upper row) and ventilation (middle row) of right upper lobe resulting in decrease of ventilation-perfusion mismatch (lower row), seen as ventilation defect (blue) with normal perfusion before therapy and normalized ventilation with normal perfusion (dark green) after therapy with ETI. Also, matching areas with combined reduced ventilation and perfusion (purple before therapy) resolved after therapy with ETI. QDP = perfusion defect percentage. (Reprinted, with permission, from reference 46.)

Esposito et al (47) prospectively evaluated the subharmonic-aided pressure estimation technique using Sonazoid microbubbles to noninvasively measure right ventricular systolic and LV diastolic pressures in 65 participants. These noninvasive measurements were comparable with measurements obtained using the reference fluid-filled pressure catheter technique, with mean errors ± SD of 1.6 mm Hg ± 1.5 (P = .85) for right ventricular systolic and 7.4 mm Hg ± 5.7 (P = .09) for LV end-diastolic pressures. Further study and validation are needed before integration into clinical workflows.

Conclusion

Continuous advancements in cardiothoracic imaging improve diagnostic performance and risk-stratification across a variety of cardiovascular and thoracic conditions.